* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
@ 2018-06-15 7:11 NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 18/24] lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c NeilBrown
` (24 more replies)
0 siblings, 25 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Here are a bunch more cleanups that will appear in my lustre-testing
tree shortly.
The interesting stuff is some enhancements to kbuild to let us
build modules from multiple directories.
I've used that to clean up the build of ptlrpc+ldlm and to
merge libcfs and lnet into the one module - I've left the lnds
separate.
I've haven't progressed further in lustre because I'm not sure which -
if any - modules are also used by the server.
I'd rather shared code was in a separate module to client-only.
Which parts of lustre/lustre/* are also used by server?
After that, it is all fairly simple cleanups.
I plan to submit the kbuild stuff upstream next week.
Thanks,
NeilBrown
---
NeilBrown (24):
kbuild: detect directories in components of a module.
kbuild: treat a directory listed in a composite object as foo/mod.a
kbuild: support building of per-directory mod.a
kbuild: disable KBUILD_MODNAME when building for mod.a
kbuild: Add documentation for modobj-m
lustre: build ldlm in the ldlm directory.
lustre: merge libcfs module into lnet.
lustre: use truncate_inode_page in place of truncate_complete_page
lustre: don't use spare bits in iattr.ia_valid
lustre: remove lustre_patchless_compat.h
lustre: discard current_n*groups macros.
lustre: discard LTIME_S macro
lustre: discard NO_QUOTA and QUOTA_OK
lustre: discard ext2* bit operations
lustre: centralize TIMES_SET_FLAGS
lustre: remove redefinition of module_init().
lustre: remove lustre_compat.h
lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
lustre: move tcd locking across to tracefile.c
lustre: move remainder of linux-tracefile.c to tracefile.c
lustre: fold cfs_tracefile_*_arch into their only callers.
lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
lustre: discard TCD_MAX_TYPES
Documentation/kbuild/makefiles.txt | 65 +++++
drivers/staging/lustre/lnet/Makefile | 7 -
drivers/staging/lustre/lnet/libcfs/Makefile | 7 -
drivers/staging/lustre/lnet/libcfs/debug.c | 6
.../staging/lustre/lnet/libcfs/linux-tracefile.c | 258 --------------------
drivers/staging/lustre/lnet/libcfs/module.c | 26 +-
drivers/staging/lustre/lnet/libcfs/tracefile.c | 241 +++++++++++++++++--
drivers/staging/lustre/lnet/libcfs/tracefile.h | 40 ---
drivers/staging/lustre/lnet/lnet/Makefile | 11 -
drivers/staging/lustre/lnet/lnet/module.c | 9 +
drivers/staging/lustre/lustre/fid/lproc_fid.c | 1
drivers/staging/lustre/lustre/fld/lproc_fld.c | 1
drivers/staging/lustre/lustre/include/cl_object.h | 6
.../staging/lustre/lustre/include/lustre_compat.h | 82 ------
drivers/staging/lustre/lustre/include/lustre_net.h | 1
.../lustre/include/lustre_patchless_compat.h | 68 -----
drivers/staging/lustre/lustre/include/obd.h | 7 +
.../staging/lustre/lustre/include/obd_support.h | 1
drivers/staging/lustre/lustre/ldlm/Makefile | 11 +
drivers/staging/lustre/lustre/llite/dir.c | 3
drivers/staging/lustre/lustre/llite/file.c | 36 +--
drivers/staging/lustre/lustre/llite/lcommon_cl.c | 11 -
.../staging/lustre/lustre/llite/llite_internal.h | 13 +
drivers/staging/lustre/lustre/llite/llite_lib.c | 52 ++--
drivers/staging/lustre/lustre/llite/namei.c | 16 +
drivers/staging/lustre/lustre/llite/vvp_io.c | 8 -
drivers/staging/lustre/lustre/llite/vvp_object.c | 6
drivers/staging/lustre/lustre/lmv/lmv_intent.c | 6
drivers/staging/lustre/lustre/lmv/lmv_obd.c | 21 +-
drivers/staging/lustre/lustre/lov/lov_io.c | 3
drivers/staging/lustre/lustre/lov/lov_obd.c | 2
drivers/staging/lustre/lustre/lov/lov_pack.c | 1
drivers/staging/lustre/lustre/lov/lov_page.c | 1
drivers/staging/lustre/lustre/mdc/mdc_lib.c | 19 +
drivers/staging/lustre/lustre/mdc/mdc_reint.c | 4
drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 -
drivers/staging/lustre/lustre/obdclass/class_obd.c | 1
drivers/staging/lustre/lustre/obdclass/llog.c | 7 -
.../lustre/lustre/obdclass/lprocfs_status.c | 1
.../staging/lustre/lustre/obdclass/obd_config.c | 1
drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1
drivers/staging/lustre/lustre/obdclass/obdo.c | 10 -
.../staging/lustre/lustre/obdecho/echo_client.c | 1
drivers/staging/lustre/lustre/osc/lproc_osc.c | 1
drivers/staging/lustre/lustre/osc/osc_cache.c | 4
drivers/staging/lustre/lustre/osc/osc_io.c | 23 +-
drivers/staging/lustre/lustre/osc/osc_quota.c | 4
drivers/staging/lustre/lustre/osc/osc_request.c | 1
drivers/staging/lustre/lustre/ptlrpc/Makefile | 12 -
drivers/staging/lustre/lustre/ptlrpc/import.c | 1
.../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c | 1
drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c | 1
drivers/staging/lustre/lustre/ptlrpc/sec.c | 15 +
drivers/staging/lustre/lustre/ptlrpc/sec_plain.c | 1
drivers/staging/lustre/lustre/ptlrpc/service.c | 1
scripts/Makefile.build | 57 +++-
scripts/Makefile.lib | 63 +++--
57 files changed, 584 insertions(+), 685 deletions(-)
delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
--
Signature
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 01/24] kbuild: detect directories in components of a module.
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (10 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 13/24] lustre: discard NO_QUOTA and QUOTA_OK NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 22/24] lustre: fold cfs_tracefile_*_arch into their only callers NeilBrown
` (12 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
This is a first step in a larger change and so can
only be fully understood in the larger context.
This patch changes the code for extracting directories
from a list of objects to extract them from real-obj-X
instead of obj-X.
This should not cause any change in behaviour yet
as listing directories as components of an object is
not currently supported and will cause an error.
A future patch will give a useful meaning to directories
listed in componsite objects.
A consequence of this change is that any subsequent use of obj-y or
obj-m will still have directories listed in it.
There are no subsequent uses of obj-y and only 2 of obj-m.
1/ obj-m is included as a dependency of __build. subdir-ym, which
contains all the directories mentioned in obj-m, is also a
dependency, so this won't change the set of final dependencies.
2/ Any rule that builds a directory listed in obj-m will find that
quite_modtag has the value '[M]'. As quiet_modtag is not used
when descending into directories, this is of no consequence.
Signed-off-by: NeilBrown <neilb@suse.com>
---
scripts/Makefile.lib | 38 +++++++++++++++++++-------------------
1 file changed, 19 insertions(+), 19 deletions(-)
diff --git a/scripts/Makefile.lib b/scripts/Makefile.lib
index 5af34a2b0cd9..2f37262b9333 100644
--- a/scripts/Makefile.lib
+++ b/scripts/Makefile.lib
@@ -25,36 +25,36 @@ lib-y := $(filter-out $(obj-y), $(sort $(lib-y) $(lib-m)))
# and -m subdirs. Just put -y's first.
modorder := $(patsubst %/,%/modules.order, $(filter %/, $(obj-y)) $(obj-m:.o=.ko))
+# if $(foo-objs), $(foo-y), or $(foo-m) exists, foo.o is a composite object
+multi-used-y := $(sort $(foreach m,$(obj-y), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y))), $(m))))
+multi-used-m := $(sort $(foreach m,$(obj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))), $(m))))
+multi-used := $(multi-used-y) $(multi-used-m)
+single-used-m := $(sort $(filter-out $(multi-used-m),$(obj-m)))
+
+# Replace multi-part objects by their individual parts,
+# including built-in.a from subdirectories
+real-obj-y := $(foreach m, $(obj-y), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y))),$($(m:.o=-objs)) $($(m:.o=-y)),$(m)))
+real-obj-m := $(foreach m, $(obj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))),$($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m)),$(m)))
+
# Handle objects in subdirs
# ---------------------------------------------------------------------------
-# o if we encounter foo/ in $(obj-y), replace it by foo/built-in.a
+# o if we encounter foo/ in $(real-obj-y), replace it by foo/built-in.a
# and add the directory to the list of dirs to descend into: $(subdir-y)
-# o if we encounter foo/ in $(obj-m), remove it from $(obj-m)
+# o if we encounter foo/ in $(real-obj-m), remove it from $(real-obj-m)
# and add the directory to the list of dirs to descend into: $(subdir-m)
-__subdir-y := $(patsubst %/,%,$(filter %/, $(obj-y)))
+__subdir-y := $(patsubst %/,%,$(filter %/, $(real-obj-y)))
subdir-y += $(__subdir-y)
-__subdir-m := $(patsubst %/,%,$(filter %/, $(obj-m)))
+__subdir-m := $(patsubst %/,%,$(filter %/, $(real-obj-m)))
subdir-m += $(__subdir-m)
-obj-y := $(patsubst %/, %/built-in.a, $(obj-y))
-obj-m := $(filter-out %/, $(obj-m))
+real-obj-y := $(patsubst %/, %/built-in.a, $(real-obj-y))
+real-obj-m := $(filter-out %/, $(real-obj-m))
# Subdirectories we need to descend into
subdir-ym := $(sort $(subdir-y) $(subdir-m))
-# if $(foo-objs), $(foo-y), or $(foo-m) exists, foo.o is a composite object
-multi-used-y := $(sort $(foreach m,$(obj-y), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y))), $(m))))
-multi-used-m := $(sort $(foreach m,$(obj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))), $(m))))
-multi-used := $(multi-used-y) $(multi-used-m)
-single-used-m := $(sort $(filter-out $(multi-used-m),$(obj-m)))
-
-# $(subdir-obj-y) is the list of objects in $(obj-y) which uses dir/ to
+# $(subdir-obj-y) is the list of objects in $(real-obj-y) which uses dir/ to
# tell kbuild to descend
-subdir-obj-y := $(filter %/built-in.a, $(obj-y))
-
-# Replace multi-part objects by their individual parts,
-# including built-in.a from subdirectories
-real-obj-y := $(foreach m, $(obj-y), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y))),$($(m:.o=-objs)) $($(m:.o=-y)),$(m)))
-real-obj-m := $(foreach m, $(obj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))),$($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m)),$(m)))
+subdir-obj-y := $(filter %/built-in.a, $(real-obj-y))
# DTB
# If CONFIG_OF_ALL_DTBS is enabled, all DT blobs are built
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 02/24] kbuild: treat a directory listed in a composite object as foo/mod.a
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (19 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 19/24] lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb() NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 14/24] lustre: discard ext2* bit operations NeilBrown
` (3 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Currently a directory is not permitted in a composite
object.
This patch changes kbuild so that a directory it assumed
to mean the file "mod.a" in that directory.
The file cannot, yet, be created, so this does not yet
affect behaviour.
There are several parts to this.
1/ strip out all the directories from obj-m so that
the directories that appear in real-obj-m must be
parts of composite objects
2/ translate those directories from foo/ to foo/mod.a
at the same time that obj-y directories becomes foo/built-in.a
3/ hold list of directories needed for modules in subdir-obj-m
so that we can descend into them as required.
4/ We need a little "dance" in "Rule to link composite objects"
where we strip the mod.a back off - so we can filter against
the foo-{objs,y,m} macros, then add it back on for
declaring dependencies.
As part of this, multi_depend gains an extra argument
being the name to append to any directory.
Signed-off-by: NeilBrown <neilb@suse.com>
---
scripts/Makefile.build | 15 ++++++++-------
scripts/Makefile.lib | 19 +++++++++++++++----
2 files changed, 23 insertions(+), 11 deletions(-)
diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index 8bdb1dc4072c..d42091532021 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -412,8 +412,9 @@ endif
$(obj)/%.o: $(src)/%.S $(objtool_dep) FORCE
$(call if_changed_rule,as_o_S)
-targets += $(filter-out $(subdir-obj-y), $(real-obj-y)) $(real-obj-m) $(lib-y)
-targets += $(extra-y) $(MAKECMDGOALS) $(always)
+targets += $(filter-out $(subdir-obj-y), $(real-obj-y))
+targets += $(filter-out $(subdir-obj-m), $(real-obj-m))
+targets += $(lib-y) $(extra-y) $(MAKECMDGOALS) $(always)
# Linker scripts preprocessor (.lds.S -> .lds)
# ---------------------------------------------------------------------------
@@ -437,7 +438,7 @@ $(obj)/%.asn1.c $(obj)/%.asn1.h: $(src)/%.asn1 $(objtree)/scripts/asn1_compiler
# ---------------------------------------------------------------------------
# To build objects in subdirs, we need to descend into the directories
-$(sort $(subdir-obj-y)): $(subdir-ym) ;
+$(sort $(subdir-obj-y) $(subdir-obj-m)): $(subdir-ym) ;
#
# Rule to compile a set of .o files into one .o file
@@ -522,16 +523,16 @@ link_multi_deps = \
$(filter $(addprefix $(obj)/, \
$($(subst $(obj)/,,$(@:.o=-objs))) \
$($(subst $(obj)/,,$(@:.o=-y))) \
-$($(subst $(obj)/,,$(@:.o=-m)))), $^)
+$($(subst $(obj)/,,$(@:.o=-m)))), $(patsubst %/mod.a,%/,$^))
quiet_cmd_link_multi-m = LD [M] $@
-cmd_link_multi-m = $(LD) $(ld_flags) -r -o $@ $(link_multi_deps) $(cmd_secanalysis)
+cmd_link_multi-m = $(LD) $(ld_flags) -r -o $@ $(patsubst %/, --whole-archive %/mod.a --no-whole-archive ,$(link_multi_deps)) $(cmd_secanalysis)
$(multi-used-m): FORCE
$(call if_changed,link_multi-m)
- @{ echo $(@:.o=.ko); echo $(link_multi_deps); \
+ @{ echo $(@:.o=.ko); echo $(patsubst %/,%/mod.a,$(link_multi_deps)); \
$(cmd_undef_syms); } > $(MODVERDIR)/$(@F:.o=.mod)
-$(call multi_depend, $(multi-used-m), .o, -objs -y -m)
+$(call multi_depend, $(multi-used-m), .o, -objs -y -m,mod.a)
targets += $(multi-used-m)
targets := $(filter-out $(PHONY), $(targets))
diff --git a/scripts/Makefile.lib b/scripts/Makefile.lib
index 2f37262b9333..c8f643144cd6 100644
--- a/scripts/Makefile.lib
+++ b/scripts/Makefile.lib
@@ -12,6 +12,13 @@ KBUILD_CFLAGS += $(subdir-ccflags-y)
# Figure out what we need to build from the various variables
# ===========================================================================
+# Directories in obj-m only cause that directory to be descended, which
+# is exactly what happens for directories in obj-y. So move all
+# directories from obj-m to obj-y. Then we will know that any directory
+# in real-obj-m is a component of some other object.
+obj-y := $(obj-y) $(filter %/, $(obj-m))
+obj-m := $(filter-out %/, $(obj-m))
+
# When an object is listed to be built compiled-in and modular,
# only build the compiled-in version
obj-m := $(filter-out $(obj-y),$(obj-m))
@@ -40,14 +47,14 @@ real-obj-m := $(foreach m, $(obj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y))
# ---------------------------------------------------------------------------
# o if we encounter foo/ in $(real-obj-y), replace it by foo/built-in.a
# and add the directory to the list of dirs to descend into: $(subdir-y)
-# o if we encounter foo/ in $(real-obj-m), remove it from $(real-obj-m)
+# o if we encounter foo/ in $(real-obj-m), replace it by foo/mod.a
# and add the directory to the list of dirs to descend into: $(subdir-m)
__subdir-y := $(patsubst %/,%,$(filter %/, $(real-obj-y)))
subdir-y += $(__subdir-y)
__subdir-m := $(patsubst %/,%,$(filter %/, $(real-obj-m)))
subdir-m += $(__subdir-m)
real-obj-y := $(patsubst %/, %/built-in.a, $(real-obj-y))
-real-obj-m := $(filter-out %/, $(real-obj-m))
+real-obj-m := $(patsubst %/, %/mod.a, $(real-obj-m))
# Subdirectories we need to descend into
subdir-ym := $(sort $(subdir-y) $(subdir-m))
@@ -55,6 +62,9 @@ subdir-ym := $(sort $(subdir-y) $(subdir-m))
# $(subdir-obj-y) is the list of objects in $(real-obj-y) which uses dir/ to
# tell kbuild to descend
subdir-obj-y := $(filter %/built-in.a, $(real-obj-y))
+# $(subdir-obj-m) is the list of objects in $(real-obj-m) which uses dir/ to
+# tell kbuild to descend
+subdir-obj-m := $(filter %/mod.a, $(real-obj-m))
# DTB
# If CONFIG_OF_ALL_DTBS is enabled, all DT blobs are built
@@ -70,6 +80,7 @@ modorder := $(addprefix $(obj)/,$(modorder))
obj-m := $(addprefix $(obj)/,$(obj-m))
lib-y := $(addprefix $(obj)/,$(lib-y))
subdir-obj-y := $(addprefix $(obj)/,$(subdir-obj-y))
+subdir-obj-m := $(addprefix $(obj)/,$(subdir-obj-m))
real-obj-y := $(addprefix $(obj)/,$(real-obj-y))
real-obj-m := $(addprefix $(obj)/,$(real-obj-m))
single-used-m := $(addprefix $(obj)/,$(single-used-m))
@@ -172,11 +183,11 @@ dtc_cpp_flags = -Wp,-MD,$(depfile).pre.tmp -nostdinc \
# Useful for describing the dependency of composite objects
# Usage:
-# $(call multi_depend, multi_used_targets, suffix_to_remove, suffix_to_add)
+# $(call multi_depend, multi_used_targets, suffix_to_remove, suffix_to_add, object_in_directories)
define multi_depend
$(foreach m, $(notdir $1), \
$(eval $(obj)/$m: \
- $(addprefix $(obj)/, $(foreach s, $3, $($(m:%$(strip $2)=%$(s)))))))
+ $(addprefix $(obj)/, $(patsubst %/,%/$4,$(foreach s, $3, $($(m:%$(strip $2)=%$(s))))))))
endef
# LEX
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 03/24] kbuild: support building of per-directory mod.a
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (2 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 16/24] lustre: remove redefinition of module_init() NeilBrown
` (20 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
This patch allows a "mod.a" to be built in any
directory. A previous patch allows that mod.a
to be included in any module or another mod.a.
This is achieved via a new pair of macros: modobj-y and modobj-m.
Anything in modobj-y is added to obj-y and is otherwise
ignored.
Anything listed in modobj-m is built, almost as though "modobj.o" was
a requested target. The objects are then combined
into mod.a.
These objects are always built with part-of-module=y.
This is sufficient to build a module from source in
multiple directories. Each "other" directory lists
something like
modobj-$(CONFIG_FOO) += bar.o bar.o bat.o
and the main directory lists
obj-$(CONFIG-FOO) = foo.o
foo-y = other1/ ../friend/other2/ module.o
Signed-off-by: NeilBrown <neilb@suse.com>
---
scripts/Makefile.build | 44 +++++++++++++++++++++++++++++++++-----------
scripts/Makefile.lib | 14 ++++++++++----
2 files changed, 43 insertions(+), 15 deletions(-)
diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index d42091532021..703bc210f71f 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -15,6 +15,8 @@ obj-y :=
obj-m :=
lib-y :=
lib-m :=
+modobj-y :=
+modobj-m :=
always :=
targets :=
subdir-y :=
@@ -80,12 +82,16 @@ ifneq ($(strip $(real-obj-y) $(need-builtin)),)
builtin-target := $(obj)/built-in.a
endif
+ifneq ($(strip $(modobj-m)),)
+modobj-target := $(obj)/mod.a
+endif
+
modorder-target := $(obj)/modules.order
# We keep a list of all modules in $(MODVERDIR)
__build: $(if $(KBUILD_BUILTIN),$(builtin-target) $(lib-target) $(extra-y)) \
- $(if $(KBUILD_MODULES),$(obj-m) $(modorder-target)) \
+ $(if $(KBUILD_MODULES),$(obj-m) $(modorder-target) $(modobj-target)) \
$(subdir-ym) $(always)
@:
@@ -119,17 +125,19 @@ modkern_cflags = \
$(KBUILD_CFLAGS_KERNEL) $(CFLAGS_KERNEL))
quiet_modtag := $(empty) $(empty)
-$(real-obj-m) : part-of-module := y
-$(real-obj-m:.o=.i) : part-of-module := y
-$(real-obj-m:.o=.s) : part-of-module := y
-$(real-obj-m:.o=.lst): part-of-module := y
+_mod_obj = $(real-obj-m) $(real-modobj-m)
+
+$(_mod_obj) : part-of-module := y
+$(_mod_obj:.o=.i) : part-of-module := y
+$(_mod_obj:.o=.s) : part-of-module := y
+$(_mod_obj:.o=.lst): part-of-module := y
-$(real-obj-m) : quiet_modtag := [M]
-$(real-obj-m:.o=.i) : quiet_modtag := [M]
-$(real-obj-m:.o=.s) : quiet_modtag := [M]
-$(real-obj-m:.o=.lst): quiet_modtag := [M]
+$(_mod_obj) : quiet_modtag := [M]
+$(_mod_obj:.o=.i) : quiet_modtag := [M]
+$(_mod_obj:.o=.s) : quiet_modtag := [M]
+$(_mod_obj:.o=.lst): quiet_modtag := [M]
-$(obj-m) : quiet_modtag := [M]
+$(obj-m) : quiet_modtag := [M]
quiet_cmd_cc_s_c = CC $(quiet_modtag) $@
cmd_cc_s_c = $(CC) $(c_flags) $(DISABLE_LTO) -fverbose-asm -S -o $@ $<
@@ -413,7 +421,7 @@ $(obj)/%.o: $(src)/%.S $(objtool_dep) FORCE
$(call if_changed_rule,as_o_S)
targets += $(filter-out $(subdir-obj-y), $(real-obj-y))
-targets += $(filter-out $(subdir-obj-m), $(real-obj-m))
+targets += $(filter-out $(subdir-obj-m) $(real-modobj-m), $(real-obj-m))
targets += $(lib-y) $(extra-y) $(MAKECMDGOALS) $(always)
# Linker scripts preprocessor (.lds.S -> .lds)
@@ -509,6 +517,20 @@ targets += $(obj)/lib-ksyms.o
endif
+ifdef modobj-target
+
+quiet_cmd_ar_modobj = AR $@
+ cmd_ar_modobj = rm -f $@; \
+ $(AR) rcTP$(KBUILD_ARFLAGS) $@ $(filter $(real-modobj-m), $^)
+
+$(modobj-target): $(real-modobj-m) FORCE
+ $(call if_changed,ar_modobj)
+
+targets += $(modobj-target)
+
+endif # modobj-target
+
+
#
# Rule to link composite objects
#
diff --git a/scripts/Makefile.lib b/scripts/Makefile.lib
index c8f643144cd6..90ee6e8c910f 100644
--- a/scripts/Makefile.lib
+++ b/scripts/Makefile.lib
@@ -16,12 +16,13 @@ KBUILD_CFLAGS += $(subdir-ccflags-y)
# is exactly what happens for directories in obj-y. So move all
# directories from obj-m to obj-y. Then we will know that any directory
# in real-obj-m is a component of some other object.
-obj-y := $(obj-y) $(filter %/, $(obj-m))
+obj-y := $(obj-y) $(filter %/, $(obj-m)) $(filter-out $(obj-y), $(modobj-y))
obj-m := $(filter-out %/, $(obj-m))
# When an object is listed to be built compiled-in and modular,
# only build the compiled-in version
obj-m := $(filter-out $(obj-y),$(obj-m))
+modobj-m := $(filter-out $(obj-y), $(modobj-m))
# Libraries are always collected in one lib file.
# Filter out objects already built-in
@@ -42,6 +43,7 @@ single-used-m := $(sort $(filter-out $(multi-used-m),$(obj-m)))
# including built-in.a from subdirectories
real-obj-y := $(foreach m, $(obj-y), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y))),$($(m:.o=-objs)) $($(m:.o=-y)),$(m)))
real-obj-m := $(foreach m, $(obj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))),$($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m)),$(m)))
+real-modobj-m := $(foreach m, $(modobj-m), $(if $(strip $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))),$($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m)),$(m)))
# Handle objects in subdirs
# ---------------------------------------------------------------------------
@@ -53,8 +55,11 @@ __subdir-y := $(patsubst %/,%,$(filter %/, $(real-obj-y)))
subdir-y += $(__subdir-y)
__subdir-m := $(patsubst %/,%,$(filter %/, $(real-obj-m)))
subdir-m += $(__subdir-m)
+__subdir-mo := $(patsubst %/,%,$(filter %/, $(real-modobj-m)))
+subdir-m += $(__subdir-mo)
real-obj-y := $(patsubst %/, %/built-in.a, $(real-obj-y))
real-obj-m := $(patsubst %/, %/mod.a, $(real-obj-m))
+real-modobj-m := $(patsubst %/, %/mod.a, $(real-modobj-m))
# Subdirectories we need to descend into
subdir-ym := $(sort $(subdir-y) $(subdir-m))
@@ -62,9 +67,9 @@ subdir-ym := $(sort $(subdir-y) $(subdir-m))
# $(subdir-obj-y) is the list of objects in $(real-obj-y) which uses dir/ to
# tell kbuild to descend
subdir-obj-y := $(filter %/built-in.a, $(real-obj-y))
-# $(subdir-obj-m) is the list of objects in $(real-obj-m) which uses dir/ to
-# tell kbuild to descend
-subdir-obj-m := $(filter %/mod.a, $(real-obj-m))
+# $(subdir-obj-m) is the list of objects in $(real-obj-m) and
+# $(real-modobj-m) which use dir/ to tell kbuild to descend
+subdir-obj-m := $(filter %/mod.a, $(real-obj-m) $(real-modobj-m))
# DTB
# If CONFIG_OF_ALL_DTBS is enabled, all DT blobs are built
@@ -83,6 +88,7 @@ subdir-obj-y := $(addprefix $(obj)/,$(subdir-obj-y))
subdir-obj-m := $(addprefix $(obj)/,$(subdir-obj-m))
real-obj-y := $(addprefix $(obj)/,$(real-obj-y))
real-obj-m := $(addprefix $(obj)/,$(real-obj-m))
+real-modobj-m := $(addprefix $(obj)/,$(real-modobj-m))
single-used-m := $(addprefix $(obj)/,$(single-used-m))
multi-used-m := $(addprefix $(obj)/,$(multi-used-m))
subdir-ym := $(addprefix $(obj)/,$(subdir-ym))
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 04/24] kbuild: disable KBUILD_MODNAME when building for mod.a
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (7 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 11/24] lustre: discard current_n*groups macros NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES NeilBrown
` (15 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
When building an object to be included in mod.a we
cannot know the name of the module. So don't define
KBUILD_MODNAME. This will ensure attempt to use
that macro when the module name isn't know will
trigger an error.
Signed-off-by: NeilBrown <neilb@suse.com>
---
scripts/Makefile.lib | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/scripts/Makefile.lib b/scripts/Makefile.lib
index 90ee6e8c910f..3b2edf767efa 100644
--- a/scripts/Makefile.lib
+++ b/scripts/Makefile.lib
@@ -96,7 +96,7 @@ subdir-ym := $(addprefix $(obj)/,$(subdir-ym))
# Finds the multi-part object the current object will be linked into.
# If the object belongs to two or more multi-part objects, all of them are
# concatenated with a colon separator.
-modname-multi = $(subst $(space),:,$(sort $(foreach m,$(multi-used),\
+modname-multi = $(subst $(space),:,$(sort $(foreach m,$(multi-used) modobj.o,\
$(if $(filter $*.o, $($(m:.o=-objs)) $($(m:.o=-y)) $($(m:.o=-m))),$(m:.o=)))))
modname = $(if $(modname-multi),$(modname-multi),$(basetarget))
@@ -106,7 +106,7 @@ modname = $(if $(modname-multi),$(modname-multi),$(basetarget))
# end up in (or would, if it gets compiled in)
name-fix = $(squote)$(quote)$(subst $(comma),_,$(subst -,_,$1))$(quote)$(squote)
basename_flags = -DKBUILD_BASENAME=$(call name-fix,$(basetarget))
-modname_flags = -DKBUILD_MODNAME=$(call name-fix,$(modname))
+modname_flags = $(if $(filter-out modobj,$(modname)),-DKBUILD_MODNAME=$(call name-fix,$(modname)))
orig_c_flags = $(KBUILD_CPPFLAGS) $(KBUILD_CFLAGS) \
$(ccflags-y) $(CFLAGS_$(basetarget).o)
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 05/24] kbuild: Add documentation for modobj-m
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (5 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 06/24] lustre: build ldlm in the ldlm directory NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 11/24] lustre: discard current_n*groups macros NeilBrown
` (17 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Add documentation for building modules
from multiple directories using modobj-m.
Signed-off-by: NeilBrown <neilb@suse.com>
---
Documentation/kbuild/makefiles.txt | 65 ++++++++++++++++++++++++++++++++++--
1 file changed, 61 insertions(+), 4 deletions(-)
diff --git a/Documentation/kbuild/makefiles.txt b/Documentation/kbuild/makefiles.txt
index 048fc39a6b91..985a60cf0663 100644
--- a/Documentation/kbuild/makefiles.txt
+++ b/Documentation/kbuild/makefiles.txt
@@ -225,17 +225,74 @@ more details, with real examples.
part of the composite object ext2.o if $(CONFIG_EXT2_FS_XATTR)
evaluates to 'y'.
+ If you wish to include code in some other directory into a
+ module, that directory should be configured to create a
+ mod.a target from the relevant code, as described below.
+ The list of components for the module (e.g. foo-y) should
+ then contain the name of the directory (with trailing /).
+ The mod.a from that directory will be built and included
+ into this module.
+
Note: Of course, when you are building objects into the kernel,
the syntax above will also work. So, if you have CONFIG_EXT2_FS=y,
kbuild will build an ext2.o file for you out of the individual
parts and then link this into built-in.a, as you would expect.
---- 3.4 Objects which export symbols
+--- 3.4 Loadable modules from multiple directories - modobj-m
+
+ One way to build modules with source from multiple directories
+ is to list relative path names of all source files in a
+ single Makefile.
+
+ Example:
+ btrfs-$(CONFIG_BTRFS_FS_RUN_SANITY_TESTS) += tests/free-space-tests.o \
+ tests/extent-buffer-tests.o tests/btrfs-tests.o \
+ ....
+
+ This works, but defeats the ability to build individual
+ directories, or individual target files such as assembly
+ language targets.
+
+ Example:
+ make btrfs/tests/free-space-tests.s
+
+ A more general way is to create a mod.a file using modobj-m
+
+ Example:
+ modobj-$(CONFIG_LUSTRE_FS) += fid_request.o fid_lib.o lproc_fid.o
+
+ modobj-y is treated identically to obj-y so if CONFIG_LUSTRE_FS=y
+ then these objects are built in to the kernel. If
+ CONFIG_LUSTRE_FS=m then these objects are gathered into an archive
+ called mod.a. Archives from other directories can be included
+ by listing the directories in modobj-m, so
+
+ Example:
+ modobj-$(CONFIG_LUSTRE_FS) += fid/ fld/
+
+ This will create a mod.a in the same directory as the
+ Makefile, combining the contents of mod.a from the listed
+ directories. The mod.a can instead be included into a module
+ by listing the directory with any other local object files.
+
+ Example:
+ obj-$(CONFIG_LUSTRE_FS) += lustre.o
+ lustre-y := module.o
+ lustre-y += obdclass/ ldlm/ ptlrpc/ fld/ osc/ mgc/
+
+ When CONFIG_LUSTRE_FS=m, kbuild will build lustre.ko from module.o and
+ the mod.a archives created in the listed directories.
+
+ When components of a module are built in a different directory
+ kbuild cannot determine the name of the module so the
+ KBUILD_MODNAME macro is left undefined.
+
+--- 3.5 Objects which export symbols
No special notation is required in the makefiles for
modules exporting symbols.
---- 3.5 Library file goals - lib-y
+--- 3.6 Library file goals - lib-y
Objects listed with obj-* are used for modules, or
combined in a built-in.a for that specific directory.
@@ -263,7 +320,7 @@ more details, with real examples.
Use of lib-y is normally restricted to lib/ and arch/*/lib.
---- 3.6 Descending down in directories
+--- 3.7 Descending down in directories
A Makefile is only responsible for building objects in its own
directory. Files in subdirectories should be taken care of by
@@ -290,7 +347,7 @@ more details, with real examples.
names. This allows kbuild to totally skip the directory if the
corresponding CONFIG_ option is neither 'y' nor 'm'.
---- 3.7 Compilation flags
+--- 3.8 Compilation flags
ccflags-y, asflags-y and ldflags-y
These three flags apply only to the kbuild makefile in which they
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 06/24] lustre: build ldlm in the ldlm directory.
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (4 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 16/24] lustre: remove redefinition of module_init() NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 05/24] kbuild: Add documentation for modobj-m NeilBrown
` (18 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
ldlm code is currently build from the ptlrpc directory,
with files listed in the Makefile of that directory.
This is because both need to be in the same module.
This make partial builds such as
make drivers/staging/lustre/lustre/ldlm/
ineffective.
kbuild recently gained the ability to build modules
from multiple directories. Use that ability to
allow ldlm code to be built in the correct directory.
Note that we need to set KBUILD_MODNAME for ldlm to
build correctly.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lustre/ldlm/Makefile | 11 +++++++++++
drivers/staging/lustre/lustre/ptlrpc/Makefile | 12 +++---------
2 files changed, 14 insertions(+), 9 deletions(-)
create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
diff --git a/drivers/staging/lustre/lustre/ldlm/Makefile b/drivers/staging/lustre/lustre/ldlm/Makefile
new file mode 100644
index 000000000000..413943230572
--- /dev/null
+++ b/drivers/staging/lustre/lustre/ldlm/Makefile
@@ -0,0 +1,11 @@
+# SPDX-License-Identifier: GPL-2.0
+ccflags-y += -I$(srctree)/drivers/staging/lustre/include
+ccflags-y += -I$(srctree)/drivers/staging/lustre/lustre/include
+ccflags-y += -DKBUILD_MODNAME='"ptlrpc"'
+
+modobj-$(CONFIG_LUSTRE_FS) += l_lock.o ldlm_lock.o
+modobj-$(CONFIG_LUSTRE_FS) += ldlm_resource.o ldlm_lib.o
+modobj-$(CONFIG_LUSTRE_FS) += ldlm_plain.o ldlm_extent.o
+modobj-$(CONFIG_LUSTRE_FS) += ldlm_request.o ldlm_lockd.o
+modobj-$(CONFIG_LUSTRE_FS) += ldlm_flock.o ldlm_inodebits.o
+modobj-$(CONFIG_LUSTRE_FS) += ldlm_pool.o
diff --git a/drivers/staging/lustre/lustre/ptlrpc/Makefile b/drivers/staging/lustre/lustre/ptlrpc/Makefile
index e0a482712d8e..9c27b850c52d 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/Makefile
+++ b/drivers/staging/lustre/lustre/ptlrpc/Makefile
@@ -3,20 +3,14 @@ ccflags-y += -I$(srctree)/drivers/staging/lustre/include
ccflags-y += -I$(srctree)/drivers/staging/lustre/lustre/include
obj-$(CONFIG_LUSTRE_FS) += ptlrpc.o
-LDLM := ../../lustre/ldlm/
-ldlm_objs := $(LDLM)l_lock.o $(LDLM)ldlm_lock.o
-ldlm_objs += $(LDLM)ldlm_resource.o $(LDLM)ldlm_lib.o
-ldlm_objs += $(LDLM)ldlm_plain.o $(LDLM)ldlm_extent.o
-ldlm_objs += $(LDLM)ldlm_request.o $(LDLM)ldlm_lockd.o
-ldlm_objs += $(LDLM)ldlm_flock.o $(LDLM)ldlm_inodebits.o
-ldlm_objs += $(LDLM)ldlm_pool.o
-ptlrpc_objs := client.o recover.o connection.o niobuf.o pack_generic.o
+ptlrpc_objs += ../ldlm/
+ptlrpc_objs += client.o recover.o connection.o niobuf.o pack_generic.o
ptlrpc_objs += events.o ptlrpc_module.o service.o pinger.o
ptlrpc_objs += llog_net.o llog_client.o import.o ptlrpcd.o
ptlrpc_objs += pers.o lproc_ptlrpc.o wiretest.o layout.o
ptlrpc_objs += sec.o sec_bulk.o sec_gc.o sec_config.o
ptlrpc_objs += sec_null.o sec_plain.o nrs.o nrs_fifo.o
-ptlrpc-y := $(ldlm_objs) $(ptlrpc_objs) sec_lproc.o
+ptlrpc-y := $(ptlrpc_objs) sec_lproc.o
ptlrpc-$(CONFIG_LUSTRE_TRANSLATE_ERRNOS) += errno.o
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 07/24] lustre: merge libcfs module into lnet.
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (14 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 15/24] lustre: centralize TIMES_SET_FLAGS NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid NeilBrown
` (8 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
We want to ultimately make libcfs disappear. This step
makes the module disappear - all libcfs code is now included
in the lnet module.
This mean the lnet module_init function needs to call lnet_init().
Also the module parameters have changed names so we need to update
the links from debugs.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lnet/Makefile | 7 ++++++-
drivers/staging/lustre/lnet/libcfs/Makefile | 5 ++---
drivers/staging/lustre/lnet/libcfs/module.c | 26 +++++++++-----------------
drivers/staging/lustre/lnet/lnet/Makefile | 11 +++++------
drivers/staging/lustre/lnet/lnet/module.c | 9 +++++++++
5 files changed, 31 insertions(+), 27 deletions(-)
diff --git a/drivers/staging/lustre/lnet/Makefile b/drivers/staging/lustre/lnet/Makefile
index 0a380fe88ce8..e28be81e36cf 100644
--- a/drivers/staging/lustre/lnet/Makefile
+++ b/drivers/staging/lustre/lnet/Makefile
@@ -1 +1,6 @@
-obj-$(CONFIG_LNET) += libcfs/ lnet/ klnds/ selftest/
+# SPDX-License-Identifier: GPL-2.0
+
+obj-$(CONFIG_LNET) += klnds/ selftest/
+obj-$(CONFIG_LNET) += lnet.o
+
+lnet-y = libcfs/ lnet/
diff --git a/drivers/staging/lustre/lnet/libcfs/Makefile b/drivers/staging/lustre/lnet/libcfs/Makefile
index 5b13edc72837..1b6937c3deee 100644
--- a/drivers/staging/lustre/lnet/libcfs/Makefile
+++ b/drivers/staging/lustre/lnet/libcfs/Makefile
@@ -1,8 +1,7 @@
# SPDX-License-Identifier: GPL-2.0
ccflags-y += -I$(srctree)/drivers/staging/lustre/include
ccflags-y += -I$(srctree)/drivers/staging/lustre/lustre/include
-
-obj-$(CONFIG_LNET) += libcfs.o
+ccflags-y += -DKBUILD_MODNAME='"lnet"'
libcfs-obj-y += linux-tracefile.o linux-debug.o
libcfs-obj-y += linux-crypto.o
@@ -13,4 +12,4 @@ libcfs-obj-y += libcfs_string.o hash.o
libcfs-obj-$(CONFIG_SMP) += libcfs_cpu.o
libcfs-obj-y += libcfs_mem.o libcfs_lock.o
-libcfs-objs := $(libcfs-obj-y)
+modobj-$(CONFIG_LNET) += $(libcfs-obj-y)
diff --git a/drivers/staging/lustre/lnet/libcfs/module.c b/drivers/staging/lustre/lnet/libcfs/module.c
index 02c404c6738e..b3eb0789425b 100644
--- a/drivers/staging/lustre/lnet/libcfs/module.c
+++ b/drivers/staging/lustre/lnet/libcfs/module.c
@@ -513,19 +513,19 @@ static struct ctl_table lnet_table[] = {
static const struct lnet_debugfs_symlink_def lnet_debugfs_symlinks[] = {
{ "console_ratelimit",
- "/sys/module/libcfs/parameters/libcfs_console_ratelimit"},
+ "/sys/module/lnet/parameters/libcfs_console_ratelimit"},
{ "debug_path",
- "/sys/module/libcfs/parameters/libcfs_debug_file_path"},
+ "/sys/module/lnet/parameters/libcfs_debug_file_path"},
{ "panic_on_lbug",
- "/sys/module/libcfs/parameters/libcfs_panic_on_lbug"},
+ "/sys/module/lnet/parameters/libcfs_panic_on_lbug"},
{ "libcfs_console_backoff",
- "/sys/module/libcfs/parameters/libcfs_console_backoff"},
+ "/sys/module/lnet/parameters/libcfs_console_backoff"},
{ "debug_mb",
- "/sys/module/libcfs/parameters/libcfs_debug_mb"},
+ "/sys/module/lnet/parameters/libcfs_debug_mb"},
{ "console_min_delay_centisecs",
- "/sys/module/libcfs/parameters/libcfs_console_min_delay"},
+ "/sys/module/lnet/parameters/libcfs_console_min_delay"},
{ "console_max_delay_centisecs",
- "/sys/module/libcfs/parameters/libcfs_console_max_delay"},
+ "/sys/module/lnet/parameters/libcfs_console_max_delay"},
{},
};
@@ -688,7 +688,7 @@ int libcfs_setup(void)
}
EXPORT_SYMBOL(libcfs_setup);
-static int libcfs_init(void)
+int libcfs_init(void)
{
int rc;
@@ -700,7 +700,7 @@ static int libcfs_init(void)
return rc;
}
-static void libcfs_exit(void)
+void libcfs_exit(void)
{
int rc;
@@ -720,11 +720,3 @@ static void libcfs_exit(void)
if (rc)
pr_err("LustreError: libcfs_debug_cleanup: %d\n", rc);
}
-
-MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
-MODULE_DESCRIPTION("Lustre helper library");
-MODULE_VERSION(LIBCFS_VERSION);
-MODULE_LICENSE("GPL");
-
-module_init(libcfs_init);
-module_exit(libcfs_exit);
diff --git a/drivers/staging/lustre/lnet/lnet/Makefile b/drivers/staging/lustre/lnet/lnet/Makefile
index cbbb52133591..8323d98a9a1a 100644
--- a/drivers/staging/lustre/lnet/lnet/Makefile
+++ b/drivers/staging/lustre/lnet/lnet/Makefile
@@ -1,10 +1,9 @@
# SPDX-License-Identifier: GPL-2.0
ccflags-y += -I$(srctree)/drivers/staging/lustre/include
ccflags-y += -I$(srctree)/drivers/staging/lustre/lustre/include
+ccflags-y += -DKBUILD_MODNAME='"lnet"'
-obj-$(CONFIG_LNET) += lnet.o
-
-lnet-y := api-ni.o config.o nidstrings.o net_fault.o \
- lib-me.o lib-msg.o lib-eq.o lib-md.o lib-ptl.o \
- lib-socket.o lib-move.o module.o lo.o \
- router.o router_proc.o acceptor.o peer.o
+modobj-$(CONFIG_LNET) += api-ni.o config.o nidstrings.o net_fault.o
+modobj-$(CONFIG_LNET) += lib-me.o lib-msg.o lib-eq.o lib-md.o lib-ptl.o
+modobj-$(CONFIG_LNET) += lib-socket.o lib-move.o module.o lo.o
+modobj-$(CONFIG_LNET) += router.o router_proc.o acceptor.o peer.o
diff --git a/drivers/staging/lustre/lnet/lnet/module.c b/drivers/staging/lustre/lnet/lnet/module.c
index 9d06664f0c17..52bbe4524bb2 100644
--- a/drivers/staging/lustre/lnet/lnet/module.c
+++ b/drivers/staging/lustre/lnet/lnet/module.c
@@ -36,6 +36,9 @@
#include <linux/lnet/lib-lnet.h>
#include <uapi/linux/lnet/lnet-dlc.h>
+int libcfs_init(void);
+void libcfs_exit(void);
+
static int config_on_load;
module_param(config_on_load, int, 0444);
MODULE_PARM_DESC(config_on_load, "configure network at module load");
@@ -192,6 +195,10 @@ static int __init lnet_init(void)
{
int rc;
+ rc = libcfs_init();
+ if (rc)
+ return rc;
+
mutex_init(&lnet_config_mutex);
rc = libcfs_setup();
@@ -228,12 +235,14 @@ static void __exit lnet_exit(void)
LASSERT(!rc);
lnet_lib_exit();
+ libcfs_exit();
}
MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
MODULE_DESCRIPTION("Lustre Networking layer");
MODULE_VERSION(LNET_VERSION);
MODULE_LICENSE("GPL");
+MODULE_ALIAS("libcfs");
module_init(lnet_init);
module_exit(lnet_exit);
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (12 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 22/24] lustre: fold cfs_tracefile_*_arch into their only callers NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-16 0:17 ` James Simmons
2018-06-15 7:11 ` [lustre-devel] [PATCH 15/24] lustre: centralize TIMES_SET_FLAGS NeilBrown
` (10 subsequent siblings)
24 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Carrying a local truncate_complete_page() is not good
for maintainability. Linux now used truncate_inode_page,
so use that instead.
For correct use, we need to pass the mapping to mdc_release_page()
rather than trusting page->mapping.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../lustre/include/lustre_patchless_compat.h | 16 ----------------
drivers/staging/lustre/lustre/llite/dir.c | 3 +--
.../staging/lustre/lustre/llite/llite_internal.h | 4 ++--
drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 ++++++------
4 files changed, 9 insertions(+), 26 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
index 298476ea7557..dca1e3dbd183 100644
--- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
@@ -41,22 +41,6 @@
#include <linux/hash.h>
#include <linux/pagemap.h>
-#define ll_delete_from_page_cache(page) delete_from_page_cache(page)
-
-static inline void
-truncate_complete_page(struct address_space *mapping, struct page *page)
-{
- if (page->mapping != mapping)
- return;
-
- if (PagePrivate(page))
- page->mapping->a_ops->invalidatepage(page, 0, PAGE_SIZE);
-
- cancel_dirty_page(page);
- ClearPageMappedToDisk(page);
- ll_delete_from_page_cache(page);
-}
-
#ifndef ATTR_CTIME_SET
/*
* set ATTR_CTIME_SET to a high value to avoid any risk of collision with other
diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
index 688dddf3ca47..cc5869cf9f93 100644
--- a/drivers/staging/lustre/lustre/llite/dir.c
+++ b/drivers/staging/lustre/lustre/llite/dir.c
@@ -167,8 +167,7 @@ void ll_release_page(struct inode *inode, struct page *page, bool remove)
if (remove) {
lock_page(page);
- if (likely(page->mapping))
- truncate_complete_page(page->mapping, page);
+ truncate_inode_page(inode->i_mapping, page);
unlock_page(page);
}
put_page(page);
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index c08a6e14b6d7..28cff58690d6 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -938,11 +938,11 @@ static inline void ll_invalidate_page(struct page *vmpage)
return;
/*
- * truncate_complete_page() calls
+ * truncate_inode_page() calls
* a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
*/
ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
- truncate_complete_page(mapping, vmpage);
+ truncate_inode_page(mapping, vmpage);
}
#define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
index cff31cb0a9ac..856e6a376852 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
@@ -922,12 +922,12 @@ static int mdc_getpage(struct obd_export *exp, const struct lu_fid *fid,
return 0;
}
-static void mdc_release_page(struct page *page, int remove)
+static void mdc_release_page(struct address_space *mapping,
+ struct page *page, int remove)
{
if (remove) {
lock_page(page);
- if (likely(page->mapping))
- truncate_complete_page(page->mapping, page);
+ truncate_inode_page(mapping, page);
unlock_page(page);
}
put_page(page);
@@ -981,7 +981,7 @@ static struct page *mdc_page_locate(struct address_space *mapping, __u64 *hash,
offset, *start, *end, *hash);
if (*hash > *end) {
kunmap(page);
- mdc_release_page(page, 0);
+ mdc_release_page(mapping, page, 0);
page = NULL;
} else if (*end != *start && *hash == *end) {
/*
@@ -991,7 +991,7 @@ static struct page *mdc_page_locate(struct address_space *mapping, __u64 *hash,
* fetch the page we want.
*/
kunmap(page);
- mdc_release_page(page,
+ mdc_release_page(mapping, page,
le32_to_cpu(dp->ldp_flags) & LDF_COLLIDE);
page = NULL;
}
@@ -1365,7 +1365,7 @@ static int mdc_read_page(struct obd_export *exp, struct md_op_data *op_data,
return rc;
fail:
kunmap(page);
- mdc_release_page(page, 1);
+ mdc_release_page(mapping, page, 1);
rc = -EIO;
goto out_unlock;
}
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (15 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 07/24] lustre: merge libcfs module into lnet NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-21 1:35 ` James Simmons
2018-06-15 7:11 ` [lustre-devel] [PATCH 20/24] lustre: move tcd locking across to tracefile.c NeilBrown
` (7 subsequent siblings)
24 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Using spare bits in iattr.ia_valid is not safe as the
bits might get used for some other purpose without
us noticing.
lustre currently used 4 spare bit, but they are all
defined in different places so this isn't immediately
obvious.
This patch brings all those bit definitions together
and defined a new op_data field (op_xvalid) to hold
the extra validity bits.
It also replaces sa_valid in struct cl_setattr_io
with sa_avalid and sa_xvalid. Changing the name is
helpful as sa_valid already has another use within
lustre.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lustre/include/cl_object.h | 5 +++-
.../staging/lustre/lustre/include/lustre_compat.h | 6 -----
.../lustre/include/lustre_patchless_compat.h | 8 -------
drivers/staging/lustre/lustre/include/obd.h | 7 ++++++
drivers/staging/lustre/lustre/llite/file.c | 12 ++++++----
drivers/staging/lustre/lustre/llite/lcommon_cl.c | 5 +++-
.../staging/lustre/lustre/llite/llite_internal.h | 5 +++-
drivers/staging/lustre/lustre/llite/llite_lib.c | 21 ++++++++++--------
drivers/staging/lustre/lustre/llite/vvp_io.c | 8 +++----
drivers/staging/lustre/lustre/lmv/lmv_obd.c | 5 +++-
drivers/staging/lustre/lustre/lov/lov_io.c | 3 ++-
drivers/staging/lustre/lustre/mdc/mdc_lib.c | 13 ++++++-----
drivers/staging/lustre/lustre/osc/osc_io.c | 23 ++++++++++----------
13 files changed, 63 insertions(+), 58 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/cl_object.h b/drivers/staging/lustre/lustre/include/cl_object.h
index 6f7b991be809..caa66763c70d 100644
--- a/drivers/staging/lustre/lustre/include/cl_object.h
+++ b/drivers/staging/lustre/lustre/include/cl_object.h
@@ -1773,7 +1773,8 @@ struct cl_io {
struct cl_setattr_io {
struct ost_lvb sa_attr;
unsigned int sa_attr_flags;
- unsigned int sa_valid;
+ unsigned int sa_avalid;
+ unsigned int sa_xvalid;
int sa_stripe_index;
const struct lu_fid *sa_parent_fid;
} ci_setattr;
@@ -2304,7 +2305,7 @@ static inline int cl_io_is_mkwrite(const struct cl_io *io)
static inline int cl_io_is_trunc(const struct cl_io *io)
{
return io->ci_type == CIT_SETATTR &&
- (io->u.ci_setattr.sa_valid & ATTR_SIZE);
+ (io->u.ci_setattr.sa_avalid & ATTR_SIZE);
}
struct cl_io *cl_io_top(struct cl_io *io);
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index 3c6db0d632dc..6d1a433e32c5 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -41,12 +41,6 @@
#include <lustre_patchless_compat.h>
-/*
- * set ATTR_BLOCKS to a high value to avoid any risk of collision with other
- * ATTR_* attributes (see bug 13828)
- */
-#define ATTR_BLOCKS (1 << 27)
-
#define current_ngroups current_cred()->group_info->ngroups
#define current_groups current_cred()->group_info->small_block
diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
index dca1e3dbd183..1bbe1a779ae9 100644
--- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
@@ -41,12 +41,4 @@
#include <linux/hash.h>
#include <linux/pagemap.h>
-#ifndef ATTR_CTIME_SET
-/*
- * set ATTR_CTIME_SET to a high value to avoid any risk of collision with other
- * ATTR_* attributes (see bug 13828)
- */
-#define ATTR_CTIME_SET (1 << 28)
-#endif
-
#endif /* LUSTRE_PATCHLESS_COMPAT_H */
diff --git a/drivers/staging/lustre/lustre/include/obd.h b/drivers/staging/lustre/lustre/include/obd.h
index b1907bbffb19..d38b6bc6b246 100644
--- a/drivers/staging/lustre/lustre/include/obd.h
+++ b/drivers/staging/lustre/lustre/include/obd.h
@@ -723,6 +723,7 @@ struct md_op_data {
/* iattr fields and blocks. */
struct iattr op_attr;
+ unsigned int op_xvalid; /* eXtra validity flags */
unsigned int op_attr_flags;
__u64 op_valid;
loff_t op_attr_blocks;
@@ -751,6 +752,12 @@ struct md_op_data {
__u32 op_default_stripe_offset;
};
+/* Flags for op_xvalid */
+#define OP_ATTR_CTIME_SET (1 << 0)
+#define OP_ATTR_BLOCKS (1 << 1)
+#define OP_ATTR_OWNEROVERRIDE (1 << 2)
+#define OP_ATTR_FLAGS (1 << 3)
+
struct md_callback {
int (*md_blocking_ast)(struct ldlm_lock *lock,
struct ldlm_lock_desc *desc,
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index e888ed6e74bc..e642996db53f 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -93,9 +93,10 @@ static void ll_prepare_close(struct inode *inode, struct md_op_data *op_data,
op_data->op_attr.ia_mtime = inode->i_mtime;
op_data->op_attr.ia_ctime = inode->i_ctime;
op_data->op_attr.ia_size = i_size_read(inode);
- op_data->op_attr.ia_valid |= ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET |
- ATTR_MTIME | ATTR_MTIME_SET |
- ATTR_CTIME | ATTR_CTIME_SET;
+ op_data->op_attr.ia_valid |= (ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET |
+ ATTR_MTIME | ATTR_MTIME_SET |
+ ATTR_CTIME);
+ op_data->op_xvalid |= OP_ATTR_CTIME_SET;
op_data->op_attr_blocks = inode->i_blocks;
op_data->op_attr_flags = ll_inode_to_ext_flags(inode->i_flags);
op_data->op_handle = och->och_fh;
@@ -161,7 +162,8 @@ static int ll_close_inode_openhandle(struct inode *inode,
op_data->op_bias |= MDS_HSM_RELEASE;
op_data->op_data_version = *(__u64 *)data;
op_data->op_lease_handle = och->och_lease_handle;
- op_data->op_attr.ia_valid |= ATTR_SIZE | ATTR_BLOCKS;
+ op_data->op_attr.ia_valid |= ATTR_SIZE;
+ op_data->op_xvalid |= OP_ATTR_BLOCKS;
break;
default:
@@ -1906,7 +1908,7 @@ static int ll_hsm_import(struct inode *inode, struct file *file,
inode_lock(inode);
- rc = ll_setattr_raw(file->f_path.dentry, attr, true);
+ rc = ll_setattr_raw(file->f_path.dentry, attr, 0, true);
if (rc == -ENODATA)
rc = 0;
diff --git a/drivers/staging/lustre/lustre/llite/lcommon_cl.c b/drivers/staging/lustre/lustre/llite/lcommon_cl.c
index d7ea39ce0cb2..9d1f3881f860 100644
--- a/drivers/staging/lustre/lustre/llite/lcommon_cl.c
+++ b/drivers/staging/lustre/lustre/llite/lcommon_cl.c
@@ -80,7 +80,7 @@ u16 cl_inode_fini_refcheck;
static DEFINE_MUTEX(cl_inode_fini_guard);
int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
- unsigned int attr_flags)
+ unsigned int xvalid, unsigned int attr_flags)
{
struct lu_env *env;
struct cl_io *io;
@@ -100,7 +100,8 @@ int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
io->u.ci_setattr.sa_attr.lvb_ctime = LTIME_S(attr->ia_ctime);
io->u.ci_setattr.sa_attr.lvb_size = attr->ia_size;
io->u.ci_setattr.sa_attr_flags = attr_flags;
- io->u.ci_setattr.sa_valid = attr->ia_valid;
+ io->u.ci_setattr.sa_avalid = attr->ia_valid;
+ io->u.ci_setattr.sa_xvalid = xvalid;
io->u.ci_setattr.sa_parent_fid = lu_object_fid(&obj->co_lu);
again:
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index 28cff58690d6..b750a48fd0e1 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -805,7 +805,8 @@ void ll_kill_super(struct super_block *sb);
struct inode *ll_inode_from_resource_lock(struct ldlm_lock *lock);
void ll_dir_clear_lsm_md(struct inode *inode);
void ll_clear_inode(struct inode *inode);
-int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import);
+int ll_setattr_raw(struct dentry *dentry, struct iattr *attr,
+ unsigned int xvalid, bool hsm_import);
int ll_setattr(struct dentry *de, struct iattr *attr);
int ll_statfs(struct dentry *de, struct kstatfs *sfs);
int ll_statfs_internal(struct super_block *sb, struct obd_statfs *osfs,
@@ -1330,7 +1331,7 @@ int ll_getparent(struct file *file, struct getparent __user *arg);
/* lcommon_cl.c */
int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
- unsigned int attr_flags);
+ unsigned int xvalid, unsigned int attr_flags);
extern struct lu_env *cl_inode_fini_env;
extern u16 cl_inode_fini_refcheck;
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index 36066c839160..cd0f2517a3b3 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -1429,7 +1429,8 @@ static int ll_md_setattr(struct dentry *dentry, struct md_op_data *op_data)
*
* In case of HSMimport, we only set attr on MDS.
*/
-int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
+int ll_setattr_raw(struct dentry *dentry, struct iattr *attr,
+ unsigned int xvalid, bool hsm_import)
{
struct inode *inode = d_inode(dentry);
struct ll_inode_info *lli = ll_i2info(inode);
@@ -1470,7 +1471,7 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
/* We mark all of the fields "set" so MDS/OST does not re-set them */
if (attr->ia_valid & ATTR_CTIME) {
attr->ia_ctime = current_time(inode);
- attr->ia_valid |= ATTR_CTIME_SET;
+ xvalid |= OP_ATTR_CTIME_SET;
}
if (!(attr->ia_valid & ATTR_ATIME_SET) &&
(attr->ia_valid & ATTR_ATIME)) {
@@ -1506,12 +1507,13 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
* If we are changing file size, file content is
* modified, flag it.
*/
- attr->ia_valid |= MDS_OPEN_OWNEROVERRIDE;
+ xvalid |= OP_ATTR_OWNEROVERRIDE;
op_data->op_bias |= MDS_DATA_MODIFIED;
clear_bit(LLIF_DATA_MODIFIED, &lli->lli_flags);
}
op_data->op_attr = *attr;
+ op_data->op_xvalid = xvalid;
rc = ll_md_setattr(dentry, op_data);
if (rc)
@@ -1532,7 +1534,8 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
* setting times to past, but it is necessary due to possible
* time de-synchronization between MDT inode and OST objects
*/
- rc = cl_setattr_ost(ll_i2info(inode)->lli_clob, attr, 0);
+ rc = cl_setattr_ost(ll_i2info(inode)->lli_clob,
+ attr, xvalid, 0);
}
/*
@@ -1589,10 +1592,11 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
int ll_setattr(struct dentry *de, struct iattr *attr)
{
int mode = d_inode(de)->i_mode;
+ unsigned int xvalid = 0;
if ((attr->ia_valid & (ATTR_CTIME | ATTR_SIZE | ATTR_MODE)) ==
(ATTR_CTIME | ATTR_SIZE | ATTR_MODE))
- attr->ia_valid |= MDS_OPEN_OWNEROVERRIDE;
+ xvalid |= OP_ATTR_OWNEROVERRIDE;
if (((attr->ia_valid & (ATTR_MODE | ATTR_FORCE | ATTR_SIZE)) ==
(ATTR_SIZE | ATTR_MODE)) &&
@@ -1613,7 +1617,7 @@ int ll_setattr(struct dentry *de, struct iattr *attr)
!(attr->ia_valid & ATTR_KILL_SGID))
attr->ia_valid |= ATTR_KILL_SGID;
- return ll_setattr_raw(de, attr, false);
+ return ll_setattr_raw(de, attr, xvalid, false);
}
int ll_statfs_internal(struct super_block *sb, struct obd_statfs *osfs,
@@ -1946,7 +1950,7 @@ int ll_iocontrol(struct inode *inode, struct file *file,
return PTR_ERR(op_data);
op_data->op_attr_flags = flags;
- op_data->op_attr.ia_valid |= ATTR_ATTR_FLAG;
+ op_data->op_xvalid |= OP_ATTR_FLAGS;
rc = md_setattr(sbi->ll_md_exp, op_data, NULL, 0, &req);
ll_finish_md_op_data(op_data);
ptlrpc_req_finished(req);
@@ -1963,8 +1967,7 @@ int ll_iocontrol(struct inode *inode, struct file *file,
if (!attr)
return -ENOMEM;
- attr->ia_valid = ATTR_ATTR_FLAG;
- rc = cl_setattr_ost(obj, attr, flags);
+ rc = cl_setattr_ost(obj, attr, OP_ATTR_FLAGS, flags);
kfree(attr);
return rc;
}
diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
index e7a4778e02e4..e65523f7628d 100644
--- a/drivers/staging/lustre/lustre/llite/vvp_io.c
+++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
@@ -556,7 +556,7 @@ static int vvp_io_setattr_lock(const struct lu_env *env,
if (new_size == 0)
enqflags = CEF_DISCARD_DATA;
} else {
- unsigned int valid = io->u.ci_setattr.sa_valid;
+ unsigned int valid = io->u.ci_setattr.sa_avalid;
if (!(valid & TIMES_SET_FLAGS))
return 0;
@@ -603,11 +603,11 @@ static int vvp_io_setattr_time(const struct lu_env *env,
cl_object_attr_lock(obj);
attr->cat_ctime = io->u.ci_setattr.sa_attr.lvb_ctime;
- if (io->u.ci_setattr.sa_valid & ATTR_ATIME_SET) {
+ if (io->u.ci_setattr.sa_avalid & ATTR_ATIME_SET) {
attr->cat_atime = io->u.ci_setattr.sa_attr.lvb_atime;
valid |= CAT_ATIME;
}
- if (io->u.ci_setattr.sa_valid & ATTR_MTIME_SET) {
+ if (io->u.ci_setattr.sa_avalid & ATTR_MTIME_SET) {
attr->cat_mtime = io->u.ci_setattr.sa_attr.lvb_mtime;
valid |= CAT_MTIME;
}
@@ -632,7 +632,7 @@ static int vvp_io_setattr_start(const struct lu_env *env,
inode_lock(inode);
}
- if (io->u.ci_setattr.sa_valid & TIMES_SET_FLAGS)
+ if (io->u.ci_setattr.sa_avalid & TIMES_SET_FLAGS)
return vvp_io_setattr_time(env, ios);
return 0;
diff --git a/drivers/staging/lustre/lustre/lmv/lmv_obd.c b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
index 65f94e6ecaad..9f427a7c7b03 100644
--- a/drivers/staging/lustre/lustre/lmv/lmv_obd.c
+++ b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
@@ -1983,8 +1983,9 @@ static int lmv_setattr(struct obd_export *exp, struct md_op_data *op_data,
struct lmv_obd *lmv = &obd->u.lmv;
struct lmv_tgt_desc *tgt;
- CDEBUG(D_INODE, "SETATTR for " DFID ", valid 0x%x\n",
- PFID(&op_data->op_fid1), op_data->op_attr.ia_valid);
+ CDEBUG(D_INODE, "SETATTR for " DFID ", valid 0x%x/0x%x\n",
+ PFID(&op_data->op_fid1), op_data->op_attr.ia_valid,
+ op_data->op_xvalid);
op_data->op_flags |= MF_MDC_CANCEL_FID1;
tgt = lmv_find_target(lmv, &op_data->op_fid1);
diff --git a/drivers/staging/lustre/lustre/lov/lov_io.c b/drivers/staging/lustre/lustre/lov/lov_io.c
index b823f8a21856..50982848bca4 100644
--- a/drivers/staging/lustre/lustre/lov/lov_io.c
+++ b/drivers/staging/lustre/lustre/lov/lov_io.c
@@ -77,7 +77,8 @@ static void lov_io_sub_inherit(struct cl_io *io, struct lov_io *lio,
io->u.ci_setattr.sa_attr = parent->u.ci_setattr.sa_attr;
io->u.ci_setattr.sa_attr_flags =
parent->u.ci_setattr.sa_attr_flags;
- io->u.ci_setattr.sa_valid = parent->u.ci_setattr.sa_valid;
+ io->u.ci_setattr.sa_avalid = parent->u.ci_setattr.sa_avalid;
+ io->u.ci_setattr.sa_xvalid = parent->u.ci_setattr.sa_xvalid;
io->u.ci_setattr.sa_stripe_index = stripe;
io->u.ci_setattr.sa_parent_fid =
parent->u.ci_setattr.sa_parent_fid;
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_lib.c b/drivers/staging/lustre/lustre/mdc/mdc_lib.c
index d582968987ff..ad7c13be5e3c 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_lib.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_lib.c
@@ -234,7 +234,7 @@ void mdc_open_pack(struct ptlrpc_request *req, struct md_op_data *op_data,
set_mrc_cr_flags(rec, cr_flags);
}
-static inline __u64 attr_pack(unsigned int ia_valid)
+static inline __u64 attr_pack(unsigned int ia_valid, unsigned int ia_xvalid)
{
__u64 sa_valid = 0;
@@ -258,19 +258,19 @@ static inline __u64 attr_pack(unsigned int ia_valid)
sa_valid |= MDS_ATTR_MTIME_SET;
if (ia_valid & ATTR_FORCE)
sa_valid |= MDS_ATTR_FORCE;
- if (ia_valid & ATTR_ATTR_FLAG)
+ if (ia_xvalid & OP_ATTR_FLAGS)
sa_valid |= MDS_ATTR_ATTR_FLAG;
if (ia_valid & ATTR_KILL_SUID)
sa_valid |= MDS_ATTR_KILL_SUID;
if (ia_valid & ATTR_KILL_SGID)
sa_valid |= MDS_ATTR_KILL_SGID;
- if (ia_valid & ATTR_CTIME_SET)
+ if (ia_xvalid & OP_ATTR_CTIME_SET)
sa_valid |= MDS_ATTR_CTIME_SET;
if (ia_valid & ATTR_OPEN)
sa_valid |= MDS_ATTR_FROM_OPEN;
- if (ia_valid & ATTR_BLOCKS)
+ if (ia_xvalid & OP_ATTR_BLOCKS)
sa_valid |= MDS_ATTR_BLOCKS;
- if (ia_valid & MDS_OPEN_OWNEROVERRIDE)
+ if (ia_xvalid & OP_ATTR_OWNEROVERRIDE)
/* NFSD hack (see bug 5781) */
sa_valid |= MDS_OPEN_OWNEROVERRIDE;
return sa_valid;
@@ -286,7 +286,8 @@ static void mdc_setattr_pack_rec(struct mdt_rec_setattr *rec,
rec->sa_suppgid = -1;
rec->sa_fid = op_data->op_fid1;
- rec->sa_valid = attr_pack(op_data->op_attr.ia_valid);
+ rec->sa_valid = attr_pack(op_data->op_attr.ia_valid,
+ op_data->op_xvalid);
rec->sa_mode = op_data->op_attr.ia_mode;
rec->sa_uid = from_kuid(&init_user_ns, op_data->op_attr.ia_uid);
rec->sa_gid = from_kgid(&init_user_ns, op_data->op_attr.ia_gid);
diff --git a/drivers/staging/lustre/lustre/osc/osc_io.c b/drivers/staging/lustre/lustre/osc/osc_io.c
index 67734a8ed331..955525f69958 100644
--- a/drivers/staging/lustre/lustre/osc/osc_io.c
+++ b/drivers/staging/lustre/lustre/osc/osc_io.c
@@ -499,7 +499,8 @@ static int osc_io_setattr_start(const struct lu_env *env,
struct obdo *oa = &oio->oi_oa;
struct osc_async_cbargs *cbargs = &oio->oi_cbarg;
__u64 size = io->u.ci_setattr.sa_attr.lvb_size;
- unsigned int ia_valid = io->u.ci_setattr.sa_valid;
+ unsigned int ia_avalid = io->u.ci_setattr.sa_avalid;
+ unsigned int ia_xvalid = io->u.ci_setattr.sa_xvalid;
int result = 0;
/* truncate cache dirty pages first */
@@ -514,20 +515,20 @@ static int osc_io_setattr_start(const struct lu_env *env,
struct ost_lvb *lvb = &io->u.ci_setattr.sa_attr;
unsigned int cl_valid = 0;
- if (ia_valid & ATTR_SIZE) {
+ if (ia_avalid & ATTR_SIZE) {
attr->cat_size = size;
attr->cat_kms = size;
cl_valid = CAT_SIZE | CAT_KMS;
}
- if (ia_valid & ATTR_MTIME_SET) {
+ if (ia_avalid & ATTR_MTIME_SET) {
attr->cat_mtime = lvb->lvb_mtime;
cl_valid |= CAT_MTIME;
}
- if (ia_valid & ATTR_ATIME_SET) {
+ if (ia_avalid & ATTR_ATIME_SET) {
attr->cat_atime = lvb->lvb_atime;
cl_valid |= CAT_ATIME;
}
- if (ia_valid & ATTR_CTIME_SET) {
+ if (ia_xvalid & OP_ATTR_CTIME_SET) {
attr->cat_ctime = lvb->lvb_ctime;
cl_valid |= CAT_CTIME;
}
@@ -542,19 +543,19 @@ static int osc_io_setattr_start(const struct lu_env *env,
obdo_set_parent_fid(oa, io->u.ci_setattr.sa_parent_fid);
oa->o_stripe_idx = io->u.ci_setattr.sa_stripe_index;
oa->o_valid |= OBD_MD_FLID | OBD_MD_FLGROUP;
- if (ia_valid & ATTR_CTIME) {
+ if (ia_avalid & ATTR_CTIME) {
oa->o_valid |= OBD_MD_FLCTIME;
oa->o_ctime = attr->cat_ctime;
}
- if (ia_valid & ATTR_ATIME) {
+ if (ia_avalid & ATTR_ATIME) {
oa->o_valid |= OBD_MD_FLATIME;
oa->o_atime = attr->cat_atime;
}
- if (ia_valid & ATTR_MTIME) {
+ if (ia_avalid & ATTR_MTIME) {
oa->o_valid |= OBD_MD_FLMTIME;
oa->o_mtime = attr->cat_mtime;
}
- if (ia_valid & ATTR_SIZE) {
+ if (ia_avalid & ATTR_SIZE) {
oa->o_size = size;
oa->o_blocks = OBD_OBJECT_EOF;
oa->o_valid |= OBD_MD_FLSIZE | OBD_MD_FLBLOCKS;
@@ -566,14 +567,14 @@ static int osc_io_setattr_start(const struct lu_env *env,
} else {
LASSERT(oio->oi_lockless == 0);
}
- if (ia_valid & ATTR_ATTR_FLAG) {
+ if (ia_xvalid & OP_ATTR_FLAGS) {
oa->o_flags = io->u.ci_setattr.sa_attr_flags;
oa->o_valid |= OBD_MD_FLFLAGS;
}
init_completion(&cbargs->opc_sync);
- if (ia_valid & ATTR_SIZE)
+ if (ia_avalid & ATTR_SIZE)
result = osc_punch_base(osc_export(cl2osc(obj)),
oa, osc_async_upcall,
cbargs, PTLRPCD_SET);
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 10/24] lustre: remove lustre_patchless_compat.h
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (17 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 20/24] lustre: move tcd locking across to tracefile.c NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 19/24] lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb() NeilBrown
` (5 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
This is now mostly empty. A few #includes need
to be added to other files to compensate.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lustre/fid/lproc_fid.c | 1
drivers/staging/lustre/lustre/fld/lproc_fld.c | 1
.../staging/lustre/lustre/include/lustre_compat.h | 2 -
drivers/staging/lustre/lustre/include/lustre_net.h | 1
.../lustre/include/lustre_patchless_compat.h | 44 --------------------
.../staging/lustre/lustre/llite/llite_internal.h | 1
drivers/staging/lustre/lustre/lov/lov_obd.c | 2 +
drivers/staging/lustre/lustre/lov/lov_pack.c | 1
drivers/staging/lustre/lustre/lov/lov_page.c | 1
drivers/staging/lustre/lustre/obdclass/class_obd.c | 1
.../lustre/lustre/obdclass/lprocfs_status.c | 1
.../staging/lustre/lustre/obdclass/obd_config.c | 1
.../staging/lustre/lustre/obdecho/echo_client.c | 1
drivers/staging/lustre/lustre/osc/lproc_osc.c | 1
drivers/staging/lustre/lustre/osc/osc_request.c | 1
.../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c | 1
drivers/staging/lustre/lustre/ptlrpc/sec_plain.c | 1
17 files changed, 15 insertions(+), 47 deletions(-)
delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
diff --git a/drivers/staging/lustre/lustre/fid/lproc_fid.c b/drivers/staging/lustre/lustre/fid/lproc_fid.c
index 0aabf473c9bd..aa2df68d5acd 100644
--- a/drivers/staging/lustre/lustre/fid/lproc_fid.c
+++ b/drivers/staging/lustre/lustre/fid/lproc_fid.c
@@ -40,6 +40,7 @@
#define DEBUG_SUBSYSTEM S_FID
#include <linux/module.h>
+#include <linux/uaccess.h>
#include <obd.h>
#include <obd_class.h>
diff --git a/drivers/staging/lustre/lustre/fld/lproc_fld.c b/drivers/staging/lustre/lustre/fld/lproc_fld.c
index 0bcfb26ef8aa..c85ef9ec639f 100644
--- a/drivers/staging/lustre/lustre/fld/lproc_fld.c
+++ b/drivers/staging/lustre/lustre/fld/lproc_fld.c
@@ -41,6 +41,7 @@
#define DEBUG_SUBSYSTEM S_FLD
#include <linux/module.h>
+#include <linux/uaccess.h>
#include <obd.h>
#include <obd_class.h>
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index 6d1a433e32c5..bf86b0612b50 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -39,8 +39,6 @@
#include <linux/cred.h>
#include <linux/module.h>
-#include <lustre_patchless_compat.h>
-
#define current_ngroups current_cred()->group_info->ngroups
#define current_groups current_cred()->group_info->small_block
diff --git a/drivers/staging/lustre/lustre/include/lustre_net.h b/drivers/staging/lustre/lustre/include/lustre_net.h
index 35b43a77eb18..dcad90bc97eb 100644
--- a/drivers/staging/lustre/lustre/include/lustre_net.h
+++ b/drivers/staging/lustre/lustre/include/lustre_net.h
@@ -51,6 +51,7 @@
* @{
*/
+#include <linux/mm.h>
#include <linux/uio.h>
#include <linux/libcfs/libcfs.h>
#include <uapi/linux/lnet/nidstr.h>
diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
deleted file mode 100644
index 1bbe1a779ae9..000000000000
--- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
+++ /dev/null
@@ -1,44 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 only,
- * as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- * General Public License version 2 for more details (a copy is included
- * in the LICENSE file that accompanied this code).
- *
- * You should have received a copy of the GNU General Public License
- * version 2 along with this program; If not, see
- * http://www.gnu.org/licenses/gpl-2.0.html
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- *
- * Copyright (c) 2011, 2012, Intel Corporation.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- */
-
-#ifndef LUSTRE_PATCHLESS_COMPAT_H
-#define LUSTRE_PATCHLESS_COMPAT_H
-
-#include <linux/fs.h>
-
-#include <linux/list.h>
-#include <linux/mm.h>
-#include <linux/hash.h>
-#include <linux/pagemap.h>
-
-#endif /* LUSTRE_PATCHLESS_COMPAT_H */
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index b750a48fd0e1..0ae5185a4998 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -39,7 +39,6 @@
#include <lustre_linkea.h>
/* for struct cl_lock_descr and struct cl_io */
-#include <lustre_patchless_compat.h>
#include <lustre_compat.h>
#include <cl_object.h>
#include <lustre_lmv.h>
diff --git a/drivers/staging/lustre/lustre/lov/lov_obd.c b/drivers/staging/lustre/lustre/lov/lov_obd.c
index 344ff4b20168..ee0898cbd9c9 100644
--- a/drivers/staging/lustre/lustre/lov/lov_obd.c
+++ b/drivers/staging/lustre/lustre/lov/lov_obd.c
@@ -40,6 +40,8 @@
#define DEBUG_SUBSYSTEM S_LOV
+#include <linux/uaccess.h>
+
#include <uapi/linux/lustre/lustre_idl.h>
#include <uapi/linux/lustre/lustre_ioctl.h>
diff --git a/drivers/staging/lustre/lustre/lov/lov_pack.c b/drivers/staging/lustre/lustre/lov/lov_pack.c
index b1060d02a164..98b114b8e93b 100644
--- a/drivers/staging/lustre/lustre/lov/lov_pack.c
+++ b/drivers/staging/lustre/lustre/lov/lov_pack.c
@@ -39,6 +39,7 @@
#define DEBUG_SUBSYSTEM S_LOV
+#include <linux/uaccess.h>
#include <lustre_net.h>
#include <lustre_swab.h>
#include <obd.h>
diff --git a/drivers/staging/lustre/lustre/lov/lov_page.c b/drivers/staging/lustre/lustre/lov/lov_page.c
index cfae1294d77a..f1c99a2a466e 100644
--- a/drivers/staging/lustre/lustre/lov/lov_page.c
+++ b/drivers/staging/lustre/lustre/lov/lov_page.c
@@ -38,6 +38,7 @@
#define DEBUG_SUBSYSTEM S_LOV
+#include <linux/highmem.h>
#include "lov_cl_internal.h"
/** \addtogroup lov
diff --git a/drivers/staging/lustre/lustre/obdclass/class_obd.c b/drivers/staging/lustre/lustre/obdclass/class_obd.c
index d6c46858941b..81a4c666bb69 100644
--- a/drivers/staging/lustre/lustre/obdclass/class_obd.c
+++ b/drivers/staging/lustre/lustre/obdclass/class_obd.c
@@ -36,6 +36,7 @@
#include <linux/atomic.h>
#include <linux/miscdevice.h>
#include <linux/libcfs/libcfs.h>
+#include <linux/uaccess.h>
#include <obd_support.h>
#include <obd_class.h>
diff --git a/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c b/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c
index bdbe6f52031a..a40ec42637e6 100644
--- a/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c
+++ b/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c
@@ -37,6 +37,7 @@
#define DEBUG_SUBSYSTEM S_CLASS
+#include <linux/uaccess.h>
#include <obd_class.h>
#include <lprocfs_status.h>
#include <uapi/linux/lustre/lustre_idl.h>
diff --git a/drivers/staging/lustre/lustre/obdclass/obd_config.c b/drivers/staging/lustre/lustre/obdclass/obd_config.c
index ffc1814398a5..cfcd17e679dd 100644
--- a/drivers/staging/lustre/lustre/obdclass/obd_config.c
+++ b/drivers/staging/lustre/lustre/obdclass/obd_config.c
@@ -37,6 +37,7 @@
#define DEBUG_SUBSYSTEM S_CLASS
+#include <linux/uaccess.h>
#include <linux/string.h>
#include <uapi/linux/lustre/lustre_ioctl.h>
diff --git a/drivers/staging/lustre/lustre/obdecho/echo_client.c b/drivers/staging/lustre/lustre/obdecho/echo_client.c
index b692e76e7108..484b8d6db6ef 100644
--- a/drivers/staging/lustre/lustre/obdecho/echo_client.c
+++ b/drivers/staging/lustre/lustre/obdecho/echo_client.c
@@ -33,6 +33,7 @@
#define DEBUG_SUBSYSTEM S_ECHO
+#include <linux/highmem.h>
#include <obd.h>
#include <obd_support.h>
#include <obd_class.h>
diff --git a/drivers/staging/lustre/lustre/osc/lproc_osc.c b/drivers/staging/lustre/lustre/osc/lproc_osc.c
index 6a705bc5420c..64931b97c887 100644
--- a/drivers/staging/lustre/lustre/osc/lproc_osc.c
+++ b/drivers/staging/lustre/lustre/osc/lproc_osc.c
@@ -32,6 +32,7 @@
*/
#define DEBUG_SUBSYSTEM S_CLASS
+#include <linux/uaccess.h>
#include <linux/statfs.h>
#include <obd_cksum.h>
#include <obd_class.h>
diff --git a/drivers/staging/lustre/lustre/osc/osc_request.c b/drivers/staging/lustre/lustre/osc/osc_request.c
index 61ef6c8d7a12..bcb9b91aa5e0 100644
--- a/drivers/staging/lustre/lustre/osc/osc_request.c
+++ b/drivers/staging/lustre/lustre/osc/osc_request.c
@@ -33,6 +33,7 @@
#define DEBUG_SUBSYSTEM S_OSC
+#include <linux/highmem.h>
#include <linux/libcfs/libcfs_hash.h>
#include <linux/sched/mm.h>
diff --git a/drivers/staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c b/drivers/staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c
index 0b638837f88b..a61e800cd2b5 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c
@@ -32,6 +32,7 @@
*/
#define DEBUG_SUBSYSTEM S_CLASS
+#include <linux/uaccess.h>
#include <obd_support.h>
#include <obd.h>
#include <lprocfs_status.h>
diff --git a/drivers/staging/lustre/lustre/ptlrpc/sec_plain.c b/drivers/staging/lustre/lustre/ptlrpc/sec_plain.c
index ec3d9af76b17..ead1df71a909 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/sec_plain.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/sec_plain.c
@@ -37,6 +37,7 @@
#define DEBUG_SUBSYSTEM S_SEC
+#include <linux/highmem.h>
#include <obd_support.h>
#include <obd_cksum.h>
#include <obd_class.h>
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 11/24] lustre: discard current_n*groups macros.
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (6 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 05/24] kbuild: Add documentation for modobj-m NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 04/24] kbuild: disable KBUILD_MODNAME when building for mod.a NeilBrown
` (16 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Just open-code the access required. This makes
the code clearer.
Also replace the 'memcpy' which violates typing
with a more correct for-loop.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lustre/include/lustre_compat.h | 3 ---
drivers/staging/lustre/lustre/ptlrpc/sec.c | 15 ++++++++++-----
2 files changed, 10 insertions(+), 8 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index bf86b0612b50..481eb4f31cfa 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -39,9 +39,6 @@
#include <linux/cred.h>
#include <linux/module.h>
-#define current_ngroups current_cred()->group_info->ngroups
-#define current_groups current_cred()->group_info->small_block
-
/*
* OBD need working random driver, thus all our
* initialization routines must be called after device
diff --git a/drivers/staging/lustre/lustre/ptlrpc/sec.c b/drivers/staging/lustre/lustre/ptlrpc/sec.c
index e193f3346e6f..9b60292370a7 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/sec.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/sec.c
@@ -2215,7 +2215,7 @@ int sptlrpc_current_user_desc_size(void)
{
int ngroups;
- ngroups = current_ngroups;
+ ngroups = current_cred()->group_info->ngroups;
if (ngroups > LUSTRE_MAX_GROUPS)
ngroups = LUSTRE_MAX_GROUPS;
@@ -2226,6 +2226,9 @@ EXPORT_SYMBOL(sptlrpc_current_user_desc_size);
int sptlrpc_pack_user_desc(struct lustre_msg *msg, int offset)
{
struct ptlrpc_user_desc *pud;
+ int ngroups;
+ kgid_t *gid;
+ int g;
pud = lustre_msg_buf(msg, offset, 0);
@@ -2240,10 +2243,12 @@ int sptlrpc_pack_user_desc(struct lustre_msg *msg, int offset)
pud->pud_ngroups = (msg->lm_buflens[offset] - sizeof(*pud)) / 4;
task_lock(current);
- if (pud->pud_ngroups > current_ngroups)
- pud->pud_ngroups = current_ngroups;
- memcpy(pud->pud_groups, current_cred()->group_info->gid,
- pud->pud_ngroups * sizeof(__u32));
+ ngroups = current_cred()->group_info->ngroups;
+ gid = current_cred()->group_info->gid;
+ if (pud->pud_ngroups > ngroups)
+ pud->pud_ngroups = ngroups;
+ for (g = 0; g < pud->pud_ngroups; g++)
+ pud->pud_groups[g] = from_kgid(&init_user_ns, gid[g]);
task_unlock(current);
return 0;
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 12/24] lustre: discard LTIME_S macro
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (21 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 14/24] lustre: discard ext2* bit operations NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 23/24] lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT NeilBrown
2018-06-21 1:48 ` [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction James Simmons
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Rather than using a macro, just access the required
field directly.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lustre/include/lustre_compat.h | 2 -
drivers/staging/lustre/lustre/llite/file.c | 24 +++++++++--------
drivers/staging/lustre/lustre/llite/lcommon_cl.c | 6 ++--
drivers/staging/lustre/lustre/llite/llite_lib.c | 28 ++++++++++----------
drivers/staging/lustre/lustre/llite/namei.c | 16 ++++++-----
drivers/staging/lustre/lustre/llite/vvp_object.c | 6 ++--
drivers/staging/lustre/lustre/lmv/lmv_intent.c | 6 ++--
drivers/staging/lustre/lustre/lmv/lmv_obd.c | 16 ++++++-----
drivers/staging/lustre/lustre/mdc/mdc_lib.c | 6 ++--
drivers/staging/lustre/lustre/mdc/mdc_reint.c | 4 +--
drivers/staging/lustre/lustre/obdclass/obdo.c | 10 ++++---
11 files changed, 61 insertions(+), 63 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index 481eb4f31cfa..5e412fa8efbd 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -49,8 +49,6 @@
#define module_init(a) late_initcall(a)
#endif
-#define LTIME_S(time) (time.tv_sec)
-
#ifndef QUOTA_OK
# define QUOTA_OK 0
#endif
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index e642996db53f..59b5fbc75a51 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -956,13 +956,13 @@ int ll_merge_attr(const struct lu_env *env, struct inode *inode)
/* merge timestamps the most recently obtained from mds with
* timestamps obtained from osts
*/
- LTIME_S(inode->i_atime) = lli->lli_atime;
- LTIME_S(inode->i_mtime) = lli->lli_mtime;
- LTIME_S(inode->i_ctime) = lli->lli_ctime;
+ inode->i_atime.tv_sec = lli->lli_atime;
+ inode->i_mtime.tv_sec = lli->lli_mtime;
+ inode->i_ctime.tv_sec = lli->lli_ctime;
- mtime = LTIME_S(inode->i_mtime);
- atime = LTIME_S(inode->i_atime);
- ctime = LTIME_S(inode->i_ctime);
+ mtime = inode->i_mtime.tv_sec;
+ atime = inode->i_atime.tv_sec;
+ ctime = inode->i_ctime.tv_sec;
cl_object_attr_lock(obj);
rc = cl_object_attr_get(env, obj, attr);
@@ -987,9 +987,9 @@ int ll_merge_attr(const struct lu_env *env, struct inode *inode)
inode->i_blocks = attr->cat_blocks;
- LTIME_S(inode->i_mtime) = mtime;
- LTIME_S(inode->i_atime) = atime;
- LTIME_S(inode->i_ctime) = ctime;
+ inode->i_mtime.tv_sec = mtime;
+ inode->i_atime.tv_sec = atime;
+ inode->i_ctime.tv_sec = ctime;
out_size_unlock:
ll_inode_size_unlock(inode);
@@ -2933,9 +2933,9 @@ static int ll_inode_revalidate(struct dentry *dentry, __u64 ibits)
return rc;
}
- LTIME_S(inode->i_atime) = ll_i2info(inode)->lli_atime;
- LTIME_S(inode->i_mtime) = ll_i2info(inode)->lli_mtime;
- LTIME_S(inode->i_ctime) = ll_i2info(inode)->lli_ctime;
+ inode->i_atime.tv_sec = ll_i2info(inode)->lli_atime;
+ inode->i_mtime.tv_sec = ll_i2info(inode)->lli_mtime;
+ inode->i_ctime.tv_sec = ll_i2info(inode)->lli_ctime;
} else {
struct ll_inode_info *lli = ll_i2info(inode);
diff --git a/drivers/staging/lustre/lustre/llite/lcommon_cl.c b/drivers/staging/lustre/lustre/llite/lcommon_cl.c
index 9d1f3881f860..6c9fe498dbc3 100644
--- a/drivers/staging/lustre/lustre/llite/lcommon_cl.c
+++ b/drivers/staging/lustre/lustre/llite/lcommon_cl.c
@@ -95,9 +95,9 @@ int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
io->ci_obj = obj;
io->ci_verify_layout = 1;
- io->u.ci_setattr.sa_attr.lvb_atime = LTIME_S(attr->ia_atime);
- io->u.ci_setattr.sa_attr.lvb_mtime = LTIME_S(attr->ia_mtime);
- io->u.ci_setattr.sa_attr.lvb_ctime = LTIME_S(attr->ia_ctime);
+ io->u.ci_setattr.sa_attr.lvb_atime = attr->ia_atime.tv_sec;
+ io->u.ci_setattr.sa_attr.lvb_mtime = attr->ia_mtime.tv_sec;
+ io->u.ci_setattr.sa_attr.lvb_ctime = attr->ia_ctime.tv_sec;
io->u.ci_setattr.sa_attr.lvb_size = attr->ia_size;
io->u.ci_setattr.sa_attr_flags = attr_flags;
io->u.ci_setattr.sa_avalid = attr->ia_valid;
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index cd0f2517a3b3..d72674873a70 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -1108,9 +1108,9 @@ static struct inode *ll_iget_anon_dir(struct super_block *sb,
LASSERTF(S_ISDIR(inode->i_mode), "Not slave inode " DFID "\n",
PFID(fid));
- LTIME_S(inode->i_mtime) = 0;
- LTIME_S(inode->i_atime) = 0;
- LTIME_S(inode->i_ctime) = 0;
+ inode->i_mtime.tv_sec = 0;
+ inode->i_atime.tv_sec = 0;
+ inode->i_ctime.tv_sec = 0;
inode->i_rdev = 0;
inode->i_op = &ll_dir_inode_operations;
@@ -1486,7 +1486,7 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr,
if (attr->ia_valid & (ATTR_MTIME | ATTR_CTIME))
CDEBUG(D_INODE, "setting mtime %lu, ctime %lu, now = %llu\n",
- LTIME_S(attr->ia_mtime), LTIME_S(attr->ia_ctime),
+ attr->ia_mtime.tv_sec, attr->ia_ctime.tv_sec,
(s64)ktime_get_real_seconds());
if (S_ISREG(inode->i_mode))
@@ -1759,23 +1759,23 @@ int ll_update_inode(struct inode *inode, struct lustre_md *md)
inode->i_generation = cl_fid_build_gen(&body->mbo_fid1);
if (body->mbo_valid & OBD_MD_FLATIME) {
- if (body->mbo_atime > LTIME_S(inode->i_atime))
- LTIME_S(inode->i_atime) = body->mbo_atime;
+ if (body->mbo_atime > inode->i_atime.tv_sec)
+ inode->i_atime.tv_sec = body->mbo_atime;
lli->lli_atime = body->mbo_atime;
}
if (body->mbo_valid & OBD_MD_FLMTIME) {
- if (body->mbo_mtime > LTIME_S(inode->i_mtime)) {
+ if (body->mbo_mtime > inode->i_mtime.tv_sec) {
CDEBUG(D_INODE,
"setting ino %lu mtime from %lu to %llu\n",
- inode->i_ino, LTIME_S(inode->i_mtime),
+ inode->i_ino, inode->i_mtime.tv_sec,
body->mbo_mtime);
- LTIME_S(inode->i_mtime) = body->mbo_mtime;
+ inode->i_mtime.tv_sec = body->mbo_mtime;
}
lli->lli_mtime = body->mbo_mtime;
}
if (body->mbo_valid & OBD_MD_FLCTIME) {
- if (body->mbo_ctime > LTIME_S(inode->i_ctime))
- LTIME_S(inode->i_ctime) = body->mbo_ctime;
+ if (body->mbo_ctime > inode->i_ctime.tv_sec)
+ inode->i_ctime.tv_sec = body->mbo_ctime;
lli->lli_ctime = body->mbo_ctime;
}
if (body->mbo_valid & OBD_MD_FLMODE)
@@ -1848,9 +1848,9 @@ int ll_read_inode2(struct inode *inode, void *opaque)
* it ourselves. They will be overwritten by either MDS or OST
* attributes - we just need to make sure they aren't newer.
*/
- LTIME_S(inode->i_mtime) = 0;
- LTIME_S(inode->i_atime) = 0;
- LTIME_S(inode->i_ctime) = 0;
+ inode->i_mtime.tv_sec = 0;
+ inode->i_atime.tv_sec = 0;
+ inode->i_ctime.tv_sec = 0;
inode->i_rdev = 0;
rc = ll_update_inode(inode, md);
if (rc)
diff --git a/drivers/staging/lustre/lustre/llite/namei.c b/drivers/staging/lustre/lustre/llite/namei.c
index d5f6d20afe8c..eb9dc532b353 100644
--- a/drivers/staging/lustre/lustre/llite/namei.c
+++ b/drivers/staging/lustre/lustre/llite/namei.c
@@ -266,9 +266,9 @@ int ll_md_blocking_ast(struct ldlm_lock *lock, struct ldlm_lock_desc *desc,
struct ll_inode_info *lli = ll_i2info(inode);
spin_lock(&lli->lli_lock);
- LTIME_S(inode->i_mtime) = 0;
- LTIME_S(inode->i_atime) = 0;
- LTIME_S(inode->i_ctime) = 0;
+ inode->i_mtime.tv_sec = 0;
+ inode->i_atime.tv_sec = 0;
+ inode->i_ctime.tv_sec = 0;
spin_unlock(&lli->lli_lock);
}
@@ -848,15 +848,15 @@ void ll_update_times(struct ptlrpc_request *request, struct inode *inode)
LASSERT(body);
if (body->mbo_valid & OBD_MD_FLMTIME &&
- body->mbo_mtime > LTIME_S(inode->i_mtime)) {
+ body->mbo_mtime > inode->i_mtime.tv_sec) {
CDEBUG(D_INODE, "setting fid " DFID " mtime from %lu to %llu\n",
- PFID(ll_inode2fid(inode)), LTIME_S(inode->i_mtime),
+ PFID(ll_inode2fid(inode)), inode->i_mtime.tv_sec,
body->mbo_mtime);
- LTIME_S(inode->i_mtime) = body->mbo_mtime;
+ inode->i_mtime.tv_sec = body->mbo_mtime;
}
if (body->mbo_valid & OBD_MD_FLCTIME &&
- body->mbo_ctime > LTIME_S(inode->i_ctime))
- LTIME_S(inode->i_ctime) = body->mbo_ctime;
+ body->mbo_ctime > inode->i_ctime.tv_sec)
+ inode->i_ctime.tv_sec = body->mbo_ctime;
}
static int ll_new_node(struct inode *dir, struct dentry *dentry,
diff --git a/drivers/staging/lustre/lustre/llite/vvp_object.c b/drivers/staging/lustre/lustre/llite/vvp_object.c
index b2cb51c8f7f4..301a90d6e5f5 100644
--- a/drivers/staging/lustre/lustre/llite/vvp_object.c
+++ b/drivers/staging/lustre/lustre/llite/vvp_object.c
@@ -170,9 +170,9 @@ static int vvp_object_glimpse(const struct lu_env *env,
{
struct inode *inode = vvp_object_inode(obj);
- lvb->lvb_mtime = LTIME_S(inode->i_mtime);
- lvb->lvb_atime = LTIME_S(inode->i_atime);
- lvb->lvb_ctime = LTIME_S(inode->i_ctime);
+ lvb->lvb_mtime = inode->i_mtime.tv_sec;
+ lvb->lvb_atime = inode->i_atime.tv_sec;
+ lvb->lvb_ctime = inode->i_ctime.tv_sec;
/*
* LU-417: Add dirty pages block count lest i_blocks reports 0, some
* "cp" or "tar" on remote node may think it's a completely sparse file
diff --git a/drivers/staging/lustre/lustre/lmv/lmv_intent.c b/drivers/staging/lustre/lustre/lmv/lmv_intent.c
index 1e850fdbc623..ba6410e800a7 100644
--- a/drivers/staging/lustre/lustre/lmv/lmv_intent.c
+++ b/drivers/staging/lustre/lustre/lmv/lmv_intent.c
@@ -233,9 +233,9 @@ int lmv_revalidate_slaves(struct obd_export *exp,
i_size_write(inode, body->mbo_size);
inode->i_blocks = body->mbo_blocks;
set_nlink(inode, body->mbo_nlink);
- LTIME_S(inode->i_atime) = body->mbo_atime;
- LTIME_S(inode->i_ctime) = body->mbo_ctime;
- LTIME_S(inode->i_mtime) = body->mbo_mtime;
+ inode->i_atime.tv_sec = body->mbo_atime;
+ inode->i_ctime.tv_sec = body->mbo_ctime;
+ inode->i_mtime.tv_sec = body->mbo_mtime;
}
md_set_lock_data(tgt->ltd_exp, lockh, inode, NULL);
diff --git a/drivers/staging/lustre/lustre/lmv/lmv_obd.c b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
index 9f427a7c7b03..58c9a8d503ec 100644
--- a/drivers/staging/lustre/lustre/lmv/lmv_obd.c
+++ b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
@@ -3030,8 +3030,8 @@ static int lmv_merge_attr(struct obd_export *exp,
CDEBUG(D_INFO, "" DFID " size %llu, blocks %llu nlink %u, atime %lu ctime %lu, mtime %lu.\n",
PFID(&lsm->lsm_md_oinfo[i].lmo_fid),
i_size_read(inode), (unsigned long long)inode->i_blocks,
- inode->i_nlink, LTIME_S(inode->i_atime),
- LTIME_S(inode->i_ctime), LTIME_S(inode->i_mtime));
+ inode->i_nlink, inode->i_atime.tv_sec,
+ inode->i_ctime.tv_sec, inode->i_mtime.tv_sec);
/* for slave stripe, it needs to subtract nlink for . and .. */
if (i)
@@ -3042,14 +3042,14 @@ static int lmv_merge_attr(struct obd_export *exp,
attr->cat_size += i_size_read(inode);
attr->cat_blocks += inode->i_blocks;
- if (attr->cat_atime < LTIME_S(inode->i_atime))
- attr->cat_atime = LTIME_S(inode->i_atime);
+ if (attr->cat_atime < inode->i_atime.tv_sec)
+ attr->cat_atime = inode->i_atime.tv_sec;
- if (attr->cat_ctime < LTIME_S(inode->i_ctime))
- attr->cat_ctime = LTIME_S(inode->i_ctime);
+ if (attr->cat_ctime < inode->i_ctime.tv_sec)
+ attr->cat_ctime = inode->i_ctime.tv_sec;
- if (attr->cat_mtime < LTIME_S(inode->i_mtime))
- attr->cat_mtime = LTIME_S(inode->i_mtime);
+ if (attr->cat_mtime < inode->i_mtime.tv_sec)
+ attr->cat_mtime = inode->i_mtime.tv_sec;
}
return 0;
}
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_lib.c b/drivers/staging/lustre/lustre/mdc/mdc_lib.c
index ad7c13be5e3c..9cb4d242d7de 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_lib.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_lib.c
@@ -293,9 +293,9 @@ static void mdc_setattr_pack_rec(struct mdt_rec_setattr *rec,
rec->sa_gid = from_kgid(&init_user_ns, op_data->op_attr.ia_gid);
rec->sa_size = op_data->op_attr.ia_size;
rec->sa_blocks = op_data->op_attr_blocks;
- rec->sa_atime = LTIME_S(op_data->op_attr.ia_atime);
- rec->sa_mtime = LTIME_S(op_data->op_attr.ia_mtime);
- rec->sa_ctime = LTIME_S(op_data->op_attr.ia_ctime);
+ rec->sa_atime = op_data->op_attr.ia_atime.tv_sec;
+ rec->sa_mtime = op_data->op_attr.ia_mtime.tv_sec;
+ rec->sa_ctime = op_data->op_attr.ia_ctime.tv_sec;
rec->sa_attr_flags = op_data->op_attr_flags;
if ((op_data->op_attr.ia_valid & ATTR_GID) &&
in_group_p(op_data->op_attr.ia_gid))
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_reint.c b/drivers/staging/lustre/lustre/mdc/mdc_reint.c
index e77c00df0693..b014eb235ce1 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_reint.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_reint.c
@@ -130,8 +130,8 @@ int mdc_setattr(struct obd_export *exp, struct md_op_data *op_data,
if (op_data->op_attr.ia_valid & (ATTR_MTIME | ATTR_CTIME))
CDEBUG(D_INODE, "setting mtime %ld, ctime %ld\n",
- LTIME_S(op_data->op_attr.ia_mtime),
- LTIME_S(op_data->op_attr.ia_ctime));
+ op_data->op_attr.ia_mtime.tv_sec,
+ op_data->op_attr.ia_ctime.tv_sec);
mdc_setattr_pack(req, op_data, ea, ealen);
req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER,
diff --git a/drivers/staging/lustre/lustre/obdclass/obdo.c b/drivers/staging/lustre/lustre/obdclass/obdo.c
index c4503bc36591..3bc8ca8363de 100644
--- a/drivers/staging/lustre/lustre/obdclass/obdo.c
+++ b/drivers/staging/lustre/lustre/obdclass/obdo.c
@@ -61,19 +61,19 @@ void obdo_from_inode(struct obdo *dst, struct inode *src, u32 valid)
if (valid & (OBD_MD_FLCTIME | OBD_MD_FLMTIME))
CDEBUG(D_INODE, "valid %x, new time %lu/%lu\n",
- valid, LTIME_S(src->i_mtime),
- LTIME_S(src->i_ctime));
+ valid, src->i_mtime.tv_sec,
+ src->i_ctime.tv_sec);
if (valid & OBD_MD_FLATIME) {
- dst->o_atime = LTIME_S(src->i_atime);
+ dst->o_atime = src->i_atime.tv_sec;
newvalid |= OBD_MD_FLATIME;
}
if (valid & OBD_MD_FLMTIME) {
- dst->o_mtime = LTIME_S(src->i_mtime);
+ dst->o_mtime = src->i_mtime.tv_sec;
newvalid |= OBD_MD_FLMTIME;
}
if (valid & OBD_MD_FLCTIME) {
- dst->o_ctime = LTIME_S(src->i_ctime);
+ dst->o_ctime = src->i_ctime.tv_sec;
newvalid |= OBD_MD_FLCTIME;
}
if (valid & OBD_MD_FLSIZE) {
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 13/24] lustre: discard NO_QUOTA and QUOTA_OK
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (9 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 01/24] kbuild: detect directories in components of a module NeilBrown
` (13 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
They don't help readability.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lustre/include/lustre_compat.h | 7 -------
drivers/staging/lustre/lustre/osc/osc_cache.c | 4 ++--
drivers/staging/lustre/lustre/osc/osc_quota.c | 4 ++--
3 files changed, 4 insertions(+), 11 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index 5e412fa8efbd..cb6e2aa35960 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -49,13 +49,6 @@
#define module_init(a) late_initcall(a)
#endif
-#ifndef QUOTA_OK
-# define QUOTA_OK 0
-#endif
-#ifndef NO_QUOTA
-# define NO_QUOTA (-EDQUOT)
-#endif
-
#if !defined(_ASM_GENERIC_BITOPS_EXT2_NON_ATOMIC_H_) && !defined(ext2_set_bit)
# define ext2_set_bit __test_and_set_bit_le
# define ext2_clear_bit __test_and_clear_bit_le
diff --git a/drivers/staging/lustre/lustre/osc/osc_cache.c b/drivers/staging/lustre/lustre/osc/osc_cache.c
index f26983004843..99de672fd078 100644
--- a/drivers/staging/lustre/lustre/osc/osc_cache.c
+++ b/drivers/staging/lustre/lustre/osc/osc_cache.c
@@ -2404,8 +2404,8 @@ int osc_queue_async_io(const struct lu_env *env, struct cl_io *io,
qid[USRQUOTA] = attr->cat_uid;
qid[GRPQUOTA] = attr->cat_gid;
- if (rc == 0 && osc_quota_chkdq(cli, qid) == NO_QUOTA)
- rc = -EDQUOT;
+ if (rc == 0)
+ rc = osc_quota_chkdq(cli, qid);
if (rc)
return rc;
}
diff --git a/drivers/staging/lustre/lustre/osc/osc_quota.c b/drivers/staging/lustre/lustre/osc/osc_quota.c
index 723ec2fb18bf..25d06dec949e 100644
--- a/drivers/staging/lustre/lustre/osc/osc_quota.c
+++ b/drivers/staging/lustre/lustre/osc/osc_quota.c
@@ -64,11 +64,11 @@ int osc_quota_chkdq(struct client_obd *cli, const unsigned int qid[])
*/
CDEBUG(D_QUOTA, "chkdq found noquota for %s %d\n",
type == USRQUOTA ? "user" : "grout", qid[type]);
- return NO_QUOTA;
+ return -EDQUOT;
}
}
- return QUOTA_OK;
+ return 0;
}
static void osc_quota_free(struct rcu_head *head)
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 14/24] lustre: discard ext2* bit operations
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (20 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 02/24] kbuild: treat a directory listed in a composite object as foo/mod.a NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 12/24] lustre: discard LTIME_S macro NeilBrown
` (2 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Just use the relevant *_le bit operation directly.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lustre/include/lustre_compat.h | 8 --------
drivers/staging/lustre/lustre/obdclass/llog.c | 6 +++---
2 files changed, 3 insertions(+), 11 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index cb6e2aa35960..534846f5561d 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -49,14 +49,6 @@
#define module_init(a) late_initcall(a)
#endif
-#if !defined(_ASM_GENERIC_BITOPS_EXT2_NON_ATOMIC_H_) && !defined(ext2_set_bit)
-# define ext2_set_bit __test_and_set_bit_le
-# define ext2_clear_bit __test_and_clear_bit_le
-# define ext2_test_bit test_bit_le
-# define ext2_find_first_zero_bit find_first_zero_bit_le
-# define ext2_find_next_zero_bit find_next_zero_bit_le
-#endif
-
#define TIMES_SET_FLAGS (ATTR_MTIME_SET | ATTR_ATIME_SET | ATTR_TIMES_SET)
#endif /* _LUSTRE_COMPAT_H */
diff --git a/drivers/staging/lustre/lustre/obdclass/llog.c b/drivers/staging/lustre/lustre/obdclass/llog.c
index bba84eae1e19..53e723d990e4 100644
--- a/drivers/staging/lustre/lustre/obdclass/llog.c
+++ b/drivers/staging/lustre/lustre/obdclass/llog.c
@@ -137,7 +137,7 @@ static int llog_read_header(const struct lu_env *env,
*/
len = llh->llh_hdr.lrh_len - llh->llh_bitmap_offset;
memset(LLOG_HDR_BITMAP(llh), 0, len - sizeof(llh->llh_tail));
- ext2_set_bit(0, LLOG_HDR_BITMAP(llh));
+ set_bit_le(0, LLOG_HDR_BITMAP(llh));
LLOG_HDR_TAIL(llh)->lrt_len = llh->llh_hdr.lrh_len;
LLOG_HDR_TAIL(llh)->lrt_index = llh->llh_hdr.lrh_index;
rc = 0;
@@ -264,7 +264,7 @@ static int llog_process_thread(void *arg)
/* skip records not set in bitmap */
while (index <= last_index &&
- !ext2_test_bit(index, LLOG_HDR_BITMAP(llh)))
+ !test_bit_le(index, LLOG_HDR_BITMAP(llh)))
++index;
if (index > last_index)
@@ -365,7 +365,7 @@ static int llog_process_thread(void *arg)
chunk_offset;
/* if set, process the callback on this record */
- if (ext2_test_bit(index, LLOG_HDR_BITMAP(llh))) {
+ if (test_bit_le(index, LLOG_HDR_BITMAP(llh))) {
rc = lpi->lpi_cb(lpi->lpi_env, loghandle, rec,
lpi->lpi_cbdata);
last_called_index = index;
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 15/24] lustre: centralize TIMES_SET_FLAGS
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (13 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 07/24] lustre: merge libcfs module into lnet NeilBrown
` (9 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
This macro is only used in llite, so move both definitions to
llite_internal.h.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lustre/include/lustre_compat.h | 2 --
.../staging/lustre/lustre/llite/llite_internal.h | 2 ++
drivers/staging/lustre/lustre/llite/llite_lib.c | 2 --
3 files changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index 534846f5561d..6d659b696b5f 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -49,6 +49,4 @@
#define module_init(a) late_initcall(a)
#endif
-#define TIMES_SET_FLAGS (ATTR_MTIME_SET | ATTR_ATIME_SET | ATTR_TIMES_SET)
-
#endif /* _LUSTRE_COMPAT_H */
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index 0ae5185a4998..91212e608c0b 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -69,6 +69,8 @@
#define LL_IT2STR(it) ((it) ? ldlm_it2str((it)->it_op) : "0")
#define LUSTRE_FPRIVATE(file) ((file)->private_data)
+#define TIMES_SET_FLAGS (ATTR_MTIME_SET | ATTR_ATIME_SET | ATTR_TIMES_SET)
+
struct ll_dentry_data {
struct lookup_intent *lld_it;
unsigned int lld_sa_generation;
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index d72674873a70..e6e02fe7703b 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -1353,8 +1353,6 @@ void ll_clear_inode(struct inode *inode)
cl_inode_fini(inode);
}
-#define TIMES_SET_FLAGS (ATTR_MTIME_SET | ATTR_ATIME_SET | ATTR_TIMES_SET)
-
static int ll_md_setattr(struct dentry *dentry, struct md_op_data *op_data)
{
struct lustre_md md;
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 16/24] lustre: remove redefinition of module_init().
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (3 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 03/24] kbuild: support building of per-directory mod.a NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 06/24] lustre: build ldlm in the ldlm directory NeilBrown
` (19 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
As lustre has only recently support building
as non-modules, this code must be very old.
If there is a dependency ordering, this is not
the way to handle it - random numbers should
be generated when filesystem is activated, not
when the module is loaded.
So just remove this code.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lustre/include/lustre_compat.h | 10 ----------
1 file changed, 10 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
index 6d659b696b5f..f555391d5ae3 100644
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
@@ -39,14 +39,4 @@
#include <linux/cred.h>
#include <linux/module.h>
-/*
- * OBD need working random driver, thus all our
- * initialization routines must be called after device
- * driver initialization
- */
-#ifndef MODULE
-#undef module_init
-#define module_init(a) late_initcall(a)
-#endif
-
#endif /* _LUSTRE_COMPAT_H */
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 17/24] lustre: remove lustre_compat.h
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 18/24] lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c NeilBrown
` (22 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
Only thing it provided was one include file,
so include that directly where needed.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lustre/include/cl_object.h | 1
.../staging/lustre/lustre/include/lustre_compat.h | 42 --------------------
.../staging/lustre/lustre/include/obd_support.h | 1
.../staging/lustre/lustre/llite/llite_internal.h | 1
drivers/staging/lustre/lustre/llite/llite_lib.c | 1
drivers/staging/lustre/lustre/obdclass/llog.c | 1
drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1
drivers/staging/lustre/lustre/ptlrpc/import.c | 1
drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c | 1
drivers/staging/lustre/lustre/ptlrpc/service.c | 1
10 files changed, 5 insertions(+), 46 deletions(-)
delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
diff --git a/drivers/staging/lustre/lustre/include/cl_object.h b/drivers/staging/lustre/lustre/include/cl_object.h
index caa66763c70d..1491beb2b9a9 100644
--- a/drivers/staging/lustre/lustre/include/cl_object.h
+++ b/drivers/staging/lustre/lustre/include/cl_object.h
@@ -90,7 +90,6 @@
* super-class definitions.
*/
#include <lu_object.h>
-#include <lustre_compat.h>
#include <linux/atomic.h>
#include <linux/mutex.h>
#include <linux/radix-tree.h>
diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
deleted file mode 100644
index f555391d5ae3..000000000000
--- a/drivers/staging/lustre/lustre/include/lustre_compat.h
+++ /dev/null
@@ -1,42 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 only,
- * as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- * General Public License version 2 for more details (a copy is included
- * in the LICENSE file that accompanied this code).
- *
- * You should have received a copy of the GNU General Public License
- * version 2 along with this program; If not, see
- * http://www.gnu.org/licenses/gpl-2.0.html
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- *
- * Copyright (c) 2011, 2012, Intel Corporation.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- */
-
-#ifndef _LUSTRE_COMPAT_H
-#define _LUSTRE_COMPAT_H
-
-#include <linux/fs_struct.h>
-#include <linux/namei.h>
-#include <linux/cred.h>
-#include <linux/module.h>
-
-#endif /* _LUSTRE_COMPAT_H */
diff --git a/drivers/staging/lustre/lustre/include/obd_support.h b/drivers/staging/lustre/lustre/include/obd_support.h
index 9e41633823f7..070a281ebb7d 100644
--- a/drivers/staging/lustre/lustre/include/obd_support.h
+++ b/drivers/staging/lustre/lustre/include/obd_support.h
@@ -37,7 +37,6 @@
#include <linux/slab.h>
#include <linux/sched/signal.h>
-#include <lustre_compat.h>
#include <lprocfs_status.h>
/* global variables */
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index 91212e608c0b..539ed3b6b64a 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -39,7 +39,6 @@
#include <lustre_linkea.h>
/* for struct cl_lock_descr and struct cl_io */
-#include <lustre_compat.h>
#include <cl_object.h>
#include <lustre_lmv.h>
#include <lustre_mdc.h>
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index e6e02fe7703b..a0a143bf4989 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -42,6 +42,7 @@
#include <linux/types.h>
#include <linux/mm.h>
#include <linux/random.h>
+#include <linux/fs_struct.h>
#include <uapi/linux/lustre/lustre_ioctl.h>
#include <lustre_ha.h>
diff --git a/drivers/staging/lustre/lustre/obdclass/llog.c b/drivers/staging/lustre/lustre/obdclass/llog.c
index 53e723d990e4..f59f89a76c91 100644
--- a/drivers/staging/lustre/lustre/obdclass/llog.c
+++ b/drivers/staging/lustre/lustre/obdclass/llog.c
@@ -45,6 +45,7 @@
#define DEBUG_SUBSYSTEM S_LOG
#include <linux/kthread.h>
+#include <linux/fs_struct.h>
#include <llog_swab.h>
#include <lustre_log.h>
#include <obd_class.h>
diff --git a/drivers/staging/lustre/lustre/obdclass/obd_mount.c b/drivers/staging/lustre/lustre/obdclass/obd_mount.c
index 06c38fdef7ba..232bbfa9242c 100644
--- a/drivers/staging/lustre/lustre/obdclass/obd_mount.c
+++ b/drivers/staging/lustre/lustre/obdclass/obd_mount.c
@@ -43,7 +43,6 @@
#include <linux/random.h>
#include <obd.h>
-#include <lustre_compat.h>
#include <obd_class.h>
#include <uapi/linux/lustre/lustre_idl.h>
#include <lustre_log.h>
diff --git a/drivers/staging/lustre/lustre/ptlrpc/import.c b/drivers/staging/lustre/lustre/ptlrpc/import.c
index 1a0f35dfab97..54ceac56c524 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/import.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/import.c
@@ -38,6 +38,7 @@
#define DEBUG_SUBSYSTEM S_RPC
#include <linux/kthread.h>
+#include <linux/fs_struct.h>
#include <obd_support.h>
#include <lustre_ha.h>
#include <lustre_net.h>
diff --git a/drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c b/drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c
index 531005411edf..66295b4fcdab 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c
@@ -53,6 +53,7 @@
#define DEBUG_SUBSYSTEM S_RPC
#include <linux/kthread.h>
+#include <linux/fs_struct.h>
#include <linux/libcfs/libcfs.h>
#include <linux/libcfs/libcfs_cpu.h>
#include <linux/libcfs/libcfs_string.h>
diff --git a/drivers/staging/lustre/lustre/ptlrpc/service.c b/drivers/staging/lustre/lustre/ptlrpc/service.c
index 3fd8c746f460..12442425702a 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/service.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/service.c
@@ -34,6 +34,7 @@
#define DEBUG_SUBSYSTEM S_RPC
#include <linux/kthread.h>
+#include <linux/fs_struct.h>
#include <obd_support.h>
#include <obd_class.h>
#include <lustre_net.h>
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 18/24] lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 17/24] lustre: remove lustre_compat.h NeilBrown
` (23 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
There is no value in keeping it separate.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lnet/libcfs/linux-tracefile.c | 22 ------------------
drivers/staging/lustre/lnet/libcfs/tracefile.c | 25 ++++++++++----------
drivers/staging/lustre/lnet/libcfs/tracefile.h | 5 ----
3 files changed, 13 insertions(+), 39 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
index 347138409eba..9e72220299af 100644
--- a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
@@ -47,8 +47,6 @@ static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
-static DECLARE_RWSEM(cfs_tracefile_sem);
-
int cfs_tracefile_init_arch(void)
{
int i;
@@ -109,26 +107,6 @@ void cfs_tracefile_fini_arch(void)
}
}
-void cfs_tracefile_read_lock(void)
-{
- down_read(&cfs_tracefile_sem);
-}
-
-void cfs_tracefile_read_unlock(void)
-{
- up_read(&cfs_tracefile_sem);
-}
-
-void cfs_tracefile_write_lock(void)
-{
- down_write(&cfs_tracefile_sem);
-}
-
-void cfs_tracefile_write_unlock(void)
-{
- up_write(&cfs_tracefile_sem);
-}
-
enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
{
if (in_irq())
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index 7ca562e156f0..5f319332f60b 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -59,6 +59,7 @@ static DEFINE_MUTEX(cfs_trace_thread_mutex);
static int thread_running;
static atomic_t cfs_tage_allocated = ATOMIC_INIT(0);
+static DECLARE_RWSEM(cfs_tracefile_sem);
struct page_collection {
struct list_head pc_pages;
@@ -711,7 +712,7 @@ int cfs_tracefile_dump_all_pages(char *filename)
mm_segment_t __oldfs;
int rc;
- cfs_tracefile_write_lock();
+ down_write(&cfs_tracefile_sem);
filp = filp_open(filename, O_CREAT | O_EXCL | O_WRONLY | O_LARGEFILE,
0600);
@@ -759,7 +760,7 @@ int cfs_tracefile_dump_all_pages(char *filename)
close:
filp_close(filp, NULL);
out:
- cfs_tracefile_write_unlock();
+ up_write(&cfs_tracefile_sem);
return rc;
}
@@ -873,12 +874,12 @@ int cfs_trace_daemon_command(char *str)
{
int rc = 0;
- cfs_tracefile_write_lock();
+ down_write(&cfs_tracefile_sem);
if (!strcmp(str, "stop")) {
- cfs_tracefile_write_unlock();
+ up_write(&cfs_tracefile_sem);
cfs_trace_stop_thread();
- cfs_tracefile_write_lock();
+ down_write(&cfs_tracefile_sem);
memset(cfs_tracefile, 0, sizeof(cfs_tracefile));
} else if (!strncmp(str, "size=", 5)) {
@@ -905,7 +906,7 @@ int cfs_trace_daemon_command(char *str)
cfs_trace_start_thread();
}
- cfs_tracefile_write_unlock();
+ up_write(&cfs_tracefile_sem);
return rc;
}
@@ -950,12 +951,12 @@ int cfs_trace_set_debug_mb(int mb)
mb /= num_possible_cpus();
pages = mb << (20 - PAGE_SHIFT);
- cfs_tracefile_write_lock();
+ down_write(&cfs_tracefile_sem);
cfs_tcd_for_each(tcd, i, j)
tcd->tcd_max_pages = (pages * tcd->tcd_pages_factor) / 100;
- cfs_tracefile_write_unlock();
+ up_write(&cfs_tracefile_sem);
return 0;
}
@@ -967,12 +968,12 @@ int cfs_trace_get_debug_mb(void)
struct cfs_trace_cpu_data *tcd;
int total_pages = 0;
- cfs_tracefile_read_lock();
+ down_read(&cfs_tracefile_sem);
cfs_tcd_for_each(tcd, i, j)
total_pages += tcd->tcd_max_pages;
- cfs_tracefile_read_unlock();
+ up_read(&cfs_tracefile_sem);
return (total_pages >> (20 - PAGE_SHIFT)) + 1;
}
@@ -1002,7 +1003,7 @@ static int tracefiled(void *arg)
goto end_loop;
filp = NULL;
- cfs_tracefile_read_lock();
+ down_read(&cfs_tracefile_sem);
if (cfs_tracefile[0]) {
filp = filp_open(cfs_tracefile,
O_CREAT | O_RDWR | O_LARGEFILE,
@@ -1014,7 +1015,7 @@ static int tracefiled(void *arg)
rc);
}
}
- cfs_tracefile_read_unlock();
+ up_read(&cfs_tracefile_sem);
if (!filp) {
put_pages_on_daemon_list(&pc);
__LASSERT(list_empty(&pc.pc_pages));
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
index 0608240d897f..9f6b73d3b6fd 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
@@ -65,11 +65,6 @@ void libcfs_run_debug_log_upcall(char *file);
int cfs_tracefile_init_arch(void);
void cfs_tracefile_fini_arch(void);
-void cfs_tracefile_read_lock(void);
-void cfs_tracefile_read_unlock(void);
-void cfs_tracefile_write_lock(void);
-void cfs_tracefile_write_unlock(void);
-
int cfs_tracefile_dump_all_pages(char *filename);
void cfs_trace_debug_print(void);
void cfs_trace_flush_pages(void);
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 19/24] lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (18 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 10/24] lustre: remove lustre_patchless_compat.h NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 02/24] kbuild: treat a directory listed in a composite object as foo/mod.a NeilBrown
` (4 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
This function in used twice.
In libcfs_debug_init() the usage is pointless as
the only place that set libcfs_debug_mb ensures
that it does not exceed the maximum. So checking
again can never help.
So open-code the small function into the only
other place that it is used - in cfs_trace_set_debug_mb(),
which is used to set libcfs_debug_mb.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lnet/libcfs/debug.c | 6 +++---
.../staging/lustre/lnet/libcfs/linux-tracefile.c | 7 -------
drivers/staging/lustre/lnet/libcfs/tracefile.c | 3 ++-
3 files changed, 5 insertions(+), 11 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/debug.c b/drivers/staging/lustre/lnet/libcfs/debug.c
index 06f694f6a28f..71effcf3ebe9 100644
--- a/drivers/staging/lustre/lnet/libcfs/debug.c
+++ b/drivers/staging/lustre/lnet/libcfs/debug.c
@@ -411,10 +411,10 @@ int libcfs_debug_init(unsigned long bufsize)
sizeof(libcfs_debug_file_path_arr));
}
- /* If libcfs_debug_mb is set to an invalid value or uninitialized
- * then just make the total buffers smp_num_cpus * TCD_MAX_PAGES
+ /* If libcfs_debug_mb is uninitialized then just make the
+ * total buffers smp_num_cpus * TCD_MAX_PAGES
*/
- if (max > cfs_trace_max_debug_mb() || max < num_possible_cpus()) {
+ if (max < num_possible_cpus()) {
max = TCD_MAX_PAGES;
} else {
max = max / num_possible_cpus();
diff --git a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
index 9e72220299af..64a5bc1e2db8 100644
--- a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
@@ -227,10 +227,3 @@ void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
hdr->ph_line_num, fn, len, buf);
}
}
-
-int cfs_trace_max_debug_mb(void)
-{
- int total_mb = (totalram_pages >> (20 - PAGE_SHIFT));
-
- return max(512, (total_mb * 80) / 100);
-}
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index 5f319332f60b..72321ce4c394 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -933,7 +933,8 @@ int cfs_trace_set_debug_mb(int mb)
int i;
int j;
int pages;
- int limit = cfs_trace_max_debug_mb();
+ int total_mb = (totalram_pages >> (20 - PAGE_SHIFT));
+ int limit = max(512, (total_mb * 80) / 100);
struct cfs_trace_cpu_data *tcd;
if (mb < num_possible_cpus()) {
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 20/24] lustre: move tcd locking across to tracefile.c
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (16 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 10/24] lustre: remove lustre_patchless_compat.h NeilBrown
` (6 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
No need to have this in linux-tracefile.c
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../staging/lustre/lnet/libcfs/linux-tracefile.c | 35 ------------
drivers/staging/lustre/lnet/libcfs/tracefile.c | 59 ++++++++++++++++++++
drivers/staging/lustre/lnet/libcfs/tracefile.h | 28 ---------
3 files changed, 59 insertions(+), 63 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
index 64a5bc1e2db8..3af77228cd7a 100644
--- a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
@@ -116,41 +116,6 @@ enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
return CFS_TCD_TYPE_PROC;
}
-/*
- * The walking argument indicates the locking comes from all tcd types
- * iterator and we must lock it and dissable local irqs to avoid deadlocks
- * with other interrupt locks that might be happening. See LU-1311
- * for details.
- */
-int cfs_trace_lock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
- __acquires(&tcd->tc_lock)
-{
- __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_MAX);
- if (tcd->tcd_type == CFS_TCD_TYPE_IRQ)
- spin_lock_irqsave(&tcd->tcd_lock, tcd->tcd_lock_flags);
- else if (tcd->tcd_type == CFS_TCD_TYPE_SOFTIRQ)
- spin_lock_bh(&tcd->tcd_lock);
- else if (unlikely(walking))
- spin_lock_irq(&tcd->tcd_lock);
- else
- spin_lock(&tcd->tcd_lock);
- return 1;
-}
-
-void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
- __releases(&tcd->tcd_lock)
-{
- __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_MAX);
- if (tcd->tcd_type == CFS_TCD_TYPE_IRQ)
- spin_unlock_irqrestore(&tcd->tcd_lock, tcd->tcd_lock_flags);
- else if (tcd->tcd_type == CFS_TCD_TYPE_SOFTIRQ)
- spin_unlock_bh(&tcd->tcd_lock);
- else if (unlikely(walking))
- spin_unlock_irq(&tcd->tcd_lock);
- else
- spin_unlock(&tcd->tcd_lock);
-}
-
void
cfs_set_ptldebug_header(struct ptldebug_header *header,
struct libcfs_debug_msg_data *msgdata,
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index 72321ce4c394..6d567a9ad8ea 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -109,6 +109,65 @@ struct cfs_trace_page {
static void put_pages_on_tcd_daemon_list(struct page_collection *pc,
struct cfs_trace_cpu_data *tcd);
+/* trace file lock routines */
+/*
+ * The walking argument indicates the locking comes from all tcd types
+ * iterator and we must lock it and dissable local irqs to avoid deadlocks
+ * with other interrupt locks that might be happening. See LU-1311
+ * for details.
+ */
+int cfs_trace_lock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
+ __acquires(&tcd->tc_lock)
+{
+ __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_MAX);
+ if (tcd->tcd_type == CFS_TCD_TYPE_IRQ)
+ spin_lock_irqsave(&tcd->tcd_lock, tcd->tcd_lock_flags);
+ else if (tcd->tcd_type == CFS_TCD_TYPE_SOFTIRQ)
+ spin_lock_bh(&tcd->tcd_lock);
+ else if (unlikely(walking))
+ spin_lock_irq(&tcd->tcd_lock);
+ else
+ spin_lock(&tcd->tcd_lock);
+ return 1;
+}
+
+void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
+ __releases(&tcd->tcd_lock)
+{
+ __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_MAX);
+ if (tcd->tcd_type == CFS_TCD_TYPE_IRQ)
+ spin_unlock_irqrestore(&tcd->tcd_lock, tcd->tcd_lock_flags);
+ else if (tcd->tcd_type == CFS_TCD_TYPE_SOFTIRQ)
+ spin_unlock_bh(&tcd->tcd_lock);
+ else if (unlikely(walking))
+ spin_unlock_irq(&tcd->tcd_lock);
+ else
+ spin_unlock(&tcd->tcd_lock);
+}
+
+#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
+ for (i = 0; cfs_trace_data[i] && \
+ (tcd = &(*cfs_trace_data[i])[cpu].tcd) && \
+ cfs_trace_lock_tcd(tcd, 1); cfs_trace_unlock_tcd(tcd, 1), i++)
+
+static inline struct cfs_trace_cpu_data *
+cfs_trace_get_tcd(void)
+{
+ struct cfs_trace_cpu_data *tcd =
+ &(*cfs_trace_data[cfs_trace_buf_idx_get()])[get_cpu()].tcd;
+
+ cfs_trace_lock_tcd(tcd, 0);
+
+ return tcd;
+}
+
+static inline void cfs_trace_put_tcd(struct cfs_trace_cpu_data *tcd)
+{
+ cfs_trace_unlock_tcd(tcd, 0);
+
+ put_cpu();
+}
+
static inline struct cfs_trace_page *
cfs_tage_from_list(struct list_head *list)
{
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
index 9f6b73d3b6fd..f49a9ba00864 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
@@ -49,8 +49,6 @@ enum cfs_trace_buf_type {
CFS_TCD_TYPE_MAX
};
-/* trace file lock routines */
-
#define TRACEFILE_NAME_SIZE 1024
extern char cfs_tracefile[TRACEFILE_NAME_SIZE];
extern long long cfs_tracefile_size;
@@ -195,11 +193,6 @@ extern union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS];
j < num_possible_cpus(); \
j++, (tcd) = &(*cfs_trace_data[i])[j].tcd)
-#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
- for (i = 0; cfs_trace_data[i] && \
- (tcd = &(*cfs_trace_data[i])[cpu].tcd) && \
- cfs_trace_lock_tcd(tcd, 1); cfs_trace_unlock_tcd(tcd, 1), i++)
-
void cfs_set_ptldebug_header(struct ptldebug_header *header,
struct libcfs_debug_msg_data *m,
unsigned long stack);
@@ -207,9 +200,6 @@ void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
const char *buf, int len, const char *file,
const char *fn);
-int cfs_trace_lock_tcd(struct cfs_trace_cpu_data *tcd, int walking);
-void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking);
-
extern char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
enum cfs_trace_buf_type cfs_trace_buf_idx_get(void);
@@ -222,24 +212,6 @@ cfs_trace_get_console_buffer(void)
return cfs_trace_console_buffers[i][j];
}
-static inline struct cfs_trace_cpu_data *
-cfs_trace_get_tcd(void)
-{
- struct cfs_trace_cpu_data *tcd =
- &(*cfs_trace_data[cfs_trace_buf_idx_get()])[get_cpu()].tcd;
-
- cfs_trace_lock_tcd(tcd, 0);
-
- return tcd;
-}
-
-static inline void cfs_trace_put_tcd(struct cfs_trace_cpu_data *tcd)
-{
- cfs_trace_unlock_tcd(tcd, 0);
-
- put_cpu();
-}
-
int cfs_trace_refill_stock(struct cfs_trace_cpu_data *tcd, gfp_t gfp,
struct list_head *stock);
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 18/24] lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 17/24] lustre: remove lustre_compat.h NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-21 1:37 ` James Simmons
2018-06-15 7:11 ` [lustre-devel] [PATCH 03/24] kbuild: support building of per-directory mod.a NeilBrown
` (21 subsequent siblings)
24 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
It's good to keep related code together.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lnet/libcfs/Makefile | 2
.../staging/lustre/lnet/libcfs/linux-tracefile.c | 194 --------------------
drivers/staging/lustre/lnet/libcfs/tracefile.c | 155 ++++++++++++++++
3 files changed, 156 insertions(+), 195 deletions(-)
delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
diff --git a/drivers/staging/lustre/lnet/libcfs/Makefile b/drivers/staging/lustre/lnet/libcfs/Makefile
index 1b6937c3deee..22c0e0c2e9cf 100644
--- a/drivers/staging/lustre/lnet/libcfs/Makefile
+++ b/drivers/staging/lustre/lnet/libcfs/Makefile
@@ -3,7 +3,7 @@ ccflags-y += -I$(srctree)/drivers/staging/lustre/include
ccflags-y += -I$(srctree)/drivers/staging/lustre/lustre/include
ccflags-y += -DKBUILD_MODNAME='"lnet"'
-libcfs-obj-y += linux-tracefile.o linux-debug.o
+libcfs-obj-y += linux-debug.o
libcfs-obj-y += linux-crypto.o
libcfs-obj-y += linux-crypto-adler.o
diff --git a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
deleted file mode 100644
index 3af77228cd7a..000000000000
--- a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
+++ /dev/null
@@ -1,194 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 only,
- * as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- * General Public License version 2 for more details (a copy is included
- * in the LICENSE file that accompanied this code).
- *
- * You should have received a copy of the GNU General Public License
- * version 2 along with this program; If not, see
- * http://www.gnu.org/licenses/gpl-2.0.html
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- *
- * Copyright (c) 2012, Intel Corporation.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- */
-
-#define DEBUG_SUBSYSTEM S_LNET
-#define LUSTRE_TRACEFILE_PRIVATE
-
-#include <linux/slab.h>
-#include <linux/mm.h>
-#include "tracefile.h"
-
-/* percents to share the total debug memory for each type */
-static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
- 80, /* 80% pages for CFS_TCD_TYPE_PROC */
- 10, /* 10% pages for CFS_TCD_TYPE_SOFTIRQ */
- 10 /* 10% pages for CFS_TCD_TYPE_IRQ */
-};
-
-char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
-
-int cfs_tracefile_init_arch(void)
-{
- int i;
- int j;
- struct cfs_trace_cpu_data *tcd;
-
- /* initialize trace_data */
- memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
- for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
- cfs_trace_data[i] =
- kmalloc_array(num_possible_cpus(),
- sizeof(union cfs_trace_data_union),
- GFP_KERNEL);
- if (!cfs_trace_data[i])
- goto out;
- }
-
- /* arch related info initialized */
- cfs_tcd_for_each(tcd, i, j) {
- spin_lock_init(&tcd->tcd_lock);
- tcd->tcd_pages_factor = pages_factor[i];
- tcd->tcd_type = i;
- tcd->tcd_cpu = j;
- }
-
- for (i = 0; i < num_possible_cpus(); i++)
- for (j = 0; j < 3; j++) {
- cfs_trace_console_buffers[i][j] =
- kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
- GFP_KERNEL);
-
- if (!cfs_trace_console_buffers[i][j])
- goto out;
- }
-
- return 0;
-
-out:
- cfs_tracefile_fini_arch();
- pr_err("lnet: Not enough memory\n");
- return -ENOMEM;
-}
-
-void cfs_tracefile_fini_arch(void)
-{
- int i;
- int j;
-
- for (i = 0; i < num_possible_cpus(); i++)
- for (j = 0; j < 3; j++) {
- kfree(cfs_trace_console_buffers[i][j]);
- cfs_trace_console_buffers[i][j] = NULL;
- }
-
- for (i = 0; cfs_trace_data[i]; i++) {
- kfree(cfs_trace_data[i]);
- cfs_trace_data[i] = NULL;
- }
-}
-
-enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
-{
- if (in_irq())
- return CFS_TCD_TYPE_IRQ;
- if (in_softirq())
- return CFS_TCD_TYPE_SOFTIRQ;
- return CFS_TCD_TYPE_PROC;
-}
-
-void
-cfs_set_ptldebug_header(struct ptldebug_header *header,
- struct libcfs_debug_msg_data *msgdata,
- unsigned long stack)
-{
- struct timespec64 ts;
-
- ktime_get_real_ts64(&ts);
-
- header->ph_subsys = msgdata->msg_subsys;
- header->ph_mask = msgdata->msg_mask;
- header->ph_cpu_id = smp_processor_id();
- header->ph_type = cfs_trace_buf_idx_get();
- /* y2038 safe since all user space treats this as unsigned, but
- * will overflow in 2106
- */
- header->ph_sec = (u32)ts.tv_sec;
- header->ph_usec = ts.tv_nsec / NSEC_PER_USEC;
- header->ph_stack = stack;
- header->ph_pid = current->pid;
- header->ph_line_num = msgdata->msg_line;
- header->ph_extern_pid = 0;
-}
-
-static char *
-dbghdr_to_err_string(struct ptldebug_header *hdr)
-{
- switch (hdr->ph_subsys) {
- case S_LND:
- case S_LNET:
- return "LNetError";
- default:
- return "LustreError";
- }
-}
-
-static char *
-dbghdr_to_info_string(struct ptldebug_header *hdr)
-{
- switch (hdr->ph_subsys) {
- case S_LND:
- case S_LNET:
- return "LNet";
- default:
- return "Lustre";
- }
-}
-
-void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
- const char *buf, int len, const char *file,
- const char *fn)
-{
- char *prefix = "Lustre", *ptype = NULL;
-
- if (mask & D_EMERG) {
- prefix = dbghdr_to_err_string(hdr);
- ptype = KERN_EMERG;
- } else if (mask & D_ERROR) {
- prefix = dbghdr_to_err_string(hdr);
- ptype = KERN_ERR;
- } else if (mask & D_WARNING) {
- prefix = dbghdr_to_info_string(hdr);
- ptype = KERN_WARNING;
- } else if (mask & (D_CONSOLE | libcfs_printk)) {
- prefix = dbghdr_to_info_string(hdr);
- ptype = KERN_INFO;
- }
-
- if (mask & D_CONSOLE) {
- pr_info("%s%s: %.*s", ptype, prefix, len, buf);
- } else {
- pr_info("%s%s: %d:%d:(%s:%d:%s()) %.*s", ptype, prefix,
- hdr->ph_pid, hdr->ph_extern_pid, file,
- hdr->ph_line_num, fn, len, buf);
- }
-}
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index 6d567a9ad8ea..617edc291035 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -168,6 +168,161 @@ static inline void cfs_trace_put_tcd(struct cfs_trace_cpu_data *tcd)
put_cpu();
}
+/* percents to share the total debug memory for each type */
+static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
+ 80, /* 80% pages for CFS_TCD_TYPE_PROC */
+ 10, /* 10% pages for CFS_TCD_TYPE_SOFTIRQ */
+ 10 /* 10% pages for CFS_TCD_TYPE_IRQ */
+};
+
+char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
+
+int cfs_tracefile_init_arch(void)
+{
+ int i;
+ int j;
+ struct cfs_trace_cpu_data *tcd;
+
+ /* initialize trace_data */
+ memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
+ for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
+ cfs_trace_data[i] =
+ kmalloc_array(num_possible_cpus(),
+ sizeof(union cfs_trace_data_union),
+ GFP_KERNEL);
+ if (!cfs_trace_data[i])
+ goto out;
+ }
+
+ /* arch related info initialized */
+ cfs_tcd_for_each(tcd, i, j) {
+ spin_lock_init(&tcd->tcd_lock);
+ tcd->tcd_pages_factor = pages_factor[i];
+ tcd->tcd_type = i;
+ tcd->tcd_cpu = j;
+ }
+
+ for (i = 0; i < num_possible_cpus(); i++)
+ for (j = 0; j < 3; j++) {
+ cfs_trace_console_buffers[i][j] =
+ kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
+ GFP_KERNEL);
+
+ if (!cfs_trace_console_buffers[i][j])
+ goto out;
+ }
+
+ return 0;
+
+out:
+ cfs_tracefile_fini_arch();
+ pr_err("lnet: Not enough memory\n");
+ return -ENOMEM;
+}
+
+void cfs_tracefile_fini_arch(void)
+{
+ int i;
+ int j;
+
+ for (i = 0; i < num_possible_cpus(); i++)
+ for (j = 0; j < 3; j++) {
+ kfree(cfs_trace_console_buffers[i][j]);
+ cfs_trace_console_buffers[i][j] = NULL;
+ }
+
+ for (i = 0; cfs_trace_data[i]; i++) {
+ kfree(cfs_trace_data[i]);
+ cfs_trace_data[i] = NULL;
+ }
+}
+
+enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
+{
+ if (in_irq())
+ return CFS_TCD_TYPE_IRQ;
+ if (in_softirq())
+ return CFS_TCD_TYPE_SOFTIRQ;
+ return CFS_TCD_TYPE_PROC;
+}
+
+void
+cfs_set_ptldebug_header(struct ptldebug_header *header,
+ struct libcfs_debug_msg_data *msgdata,
+ unsigned long stack)
+{
+ struct timespec64 ts;
+
+ ktime_get_real_ts64(&ts);
+
+ header->ph_subsys = msgdata->msg_subsys;
+ header->ph_mask = msgdata->msg_mask;
+ header->ph_cpu_id = smp_processor_id();
+ header->ph_type = cfs_trace_buf_idx_get();
+ /* y2038 safe since all user space treats this as unsigned, but
+ * will overflow in 2106
+ */
+ header->ph_sec = (u32)ts.tv_sec;
+ header->ph_usec = ts.tv_nsec / NSEC_PER_USEC;
+ header->ph_stack = stack;
+ header->ph_pid = current->pid;
+ header->ph_line_num = msgdata->msg_line;
+ header->ph_extern_pid = 0;
+}
+
+static char *
+dbghdr_to_err_string(struct ptldebug_header *hdr)
+{
+ switch (hdr->ph_subsys) {
+ case S_LND:
+ case S_LNET:
+ return "LNetError";
+ default:
+ return "LustreError";
+ }
+}
+
+static char *
+dbghdr_to_info_string(struct ptldebug_header *hdr)
+{
+ switch (hdr->ph_subsys) {
+ case S_LND:
+ case S_LNET:
+ return "LNet";
+ default:
+ return "Lustre";
+ }
+}
+
+void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
+ const char *buf, int len, const char *file,
+ const char *fn)
+{
+ char *prefix = "Lustre", *ptype = NULL;
+
+ if (mask & D_EMERG) {
+ prefix = dbghdr_to_err_string(hdr);
+ ptype = KERN_EMERG;
+ } else if (mask & D_ERROR) {
+ prefix = dbghdr_to_err_string(hdr);
+ ptype = KERN_ERR;
+ } else if (mask & D_WARNING) {
+ prefix = dbghdr_to_info_string(hdr);
+ ptype = KERN_WARNING;
+ } else if (mask & (D_CONSOLE | libcfs_printk)) {
+ prefix = dbghdr_to_info_string(hdr);
+ ptype = KERN_INFO;
+ }
+
+ if (mask & D_CONSOLE) {
+ pr_info("%s%s: %.*s", ptype, prefix, len, buf);
+ } else {
+ pr_info("%s%s: %d:%d:(%s:%d:%s()) %.*s", ptype, prefix,
+ hdr->ph_pid, hdr->ph_extern_pid, file,
+ hdr->ph_line_num, fn, len, buf);
+ }
+}
+
static inline struct cfs_trace_page *
cfs_tage_from_list(struct list_head *list)
{
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 22/24] lustre: fold cfs_tracefile_*_arch into their only callers.
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (11 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 01/24] kbuild: detect directories in components of a module NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page NeilBrown
` (11 subsequent siblings)
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
There is no need to separate "arch" init/fini from
the rest, so fold it all in.
This requires some slightly subtle changes to clean-up
to make sure we don't walk lists before they are
initialized.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lnet/libcfs/tracefile.c | 116 ++++++++++--------------
1 file changed, 49 insertions(+), 67 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index 617edc291035..fbd05905e1c4 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -177,65 +177,6 @@ static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
-int cfs_tracefile_init_arch(void)
-{
- int i;
- int j;
- struct cfs_trace_cpu_data *tcd;
-
- /* initialize trace_data */
- memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
- for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
- cfs_trace_data[i] =
- kmalloc_array(num_possible_cpus(),
- sizeof(union cfs_trace_data_union),
- GFP_KERNEL);
- if (!cfs_trace_data[i])
- goto out;
- }
-
- /* arch related info initialized */
- cfs_tcd_for_each(tcd, i, j) {
- spin_lock_init(&tcd->tcd_lock);
- tcd->tcd_pages_factor = pages_factor[i];
- tcd->tcd_type = i;
- tcd->tcd_cpu = j;
- }
-
- for (i = 0; i < num_possible_cpus(); i++)
- for (j = 0; j < 3; j++) {
- cfs_trace_console_buffers[i][j] =
- kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
- GFP_KERNEL);
-
- if (!cfs_trace_console_buffers[i][j])
- goto out;
- }
-
- return 0;
-
-out:
- cfs_tracefile_fini_arch();
- pr_err("lnet: Not enough memory\n");
- return -ENOMEM;
-}
-
-void cfs_tracefile_fini_arch(void)
-{
- int i;
- int j;
-
- for (i = 0; i < num_possible_cpus(); i++)
- for (j = 0; j < 3; j++) {
- kfree(cfs_trace_console_buffers[i][j]);
- cfs_trace_console_buffers[i][j] = NULL;
- }
-
- for (i = 0; cfs_trace_data[i]; i++) {
- kfree(cfs_trace_data[i]);
- cfs_trace_data[i] = NULL;
- }
-}
enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
{
@@ -1343,21 +1284,33 @@ void cfs_trace_stop_thread(void)
mutex_unlock(&cfs_trace_thread_mutex);
}
+static void cfs_trace_cleanup(void);
+
int cfs_tracefile_init(int max_pages)
{
struct cfs_trace_cpu_data *tcd;
int i;
int j;
- int rc;
- int factor;
- rc = cfs_tracefile_init_arch();
- if (rc)
- return rc;
+ /* initialize trace_data */
+ memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
+ for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
+ cfs_trace_data[i] =
+ kmalloc_array(num_possible_cpus(),
+ sizeof(union cfs_trace_data_union),
+ GFP_KERNEL);
+ if (!cfs_trace_data[i])
+ goto out;
+ }
+ /* arch related info initialized */
cfs_tcd_for_each(tcd, i, j) {
- /* tcd_pages_factor is initialized int tracefile_init_arch. */
- factor = tcd->tcd_pages_factor;
+ int factor = pages_factor[i];
+
+ spin_lock_init(&tcd->tcd_lock);
+ tcd->tcd_pages_factor = factor;
+ tcd->tcd_type = i;
+ tcd->tcd_cpu = j;
INIT_LIST_HEAD(&tcd->tcd_pages);
INIT_LIST_HEAD(&tcd->tcd_stock_pages);
INIT_LIST_HEAD(&tcd->tcd_daemon_pages);
@@ -1369,7 +1322,21 @@ int cfs_tracefile_init(int max_pages)
tcd->tcd_shutting_down = 0;
}
+ for (i = 0; i < num_possible_cpus(); i++)
+ for (j = 0; j < 3; j++) {
+ cfs_trace_console_buffers[i][j] =
+ kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
+ GFP_KERNEL);
+
+ if (!cfs_trace_console_buffers[i][j])
+ goto out;
+ }
+
return 0;
+out:
+ cfs_trace_cleanup();
+ pr_err("lnet: Not enough memory\n");
+ return -ENOMEM;
}
static void trace_cleanup_on_all_cpus(void)
@@ -1381,6 +1348,9 @@ static void trace_cleanup_on_all_cpus(void)
for_each_possible_cpu(cpu) {
cfs_tcd_for_each_type_lock(tcd, i, cpu) {
+ if (!tcd->tcd_pages_factor)
+ /* Not initialised */
+ continue;
tcd->tcd_shutting_down = 1;
list_for_each_entry_safe(tage, tmp, &tcd->tcd_pages,
@@ -1399,12 +1369,24 @@ static void trace_cleanup_on_all_cpus(void)
static void cfs_trace_cleanup(void)
{
struct page_collection pc;
+ int i;
+ int j;
INIT_LIST_HEAD(&pc.pc_pages);
trace_cleanup_on_all_cpus();
- cfs_tracefile_fini_arch();
+ for (i = 0; i < num_possible_cpus(); i++)
+ for (j = 0; j < 3; j++) {
+ kfree(cfs_trace_console_buffers[i][j]);
+ cfs_trace_console_buffers[i][j] = NULL;
+ }
+
+ for (i = 0; cfs_trace_data[i]; i++) {
+ kfree(cfs_trace_data[i]);
+ cfs_trace_data[i] = NULL;
+ }
+
}
void cfs_tracefile_exit(void)
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 23/24] lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (22 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 12/24] lustre: discard LTIME_S macro NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-21 1:48 ` [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction James Simmons
24 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
The possible TCD types are 0, 1, 2.
So the MAX is 2.
The count of the number of types is 3.
CFS_TCD_TYPE_MAX is 3 - obviously wrong.
So rename it to CFS_TCD_TYPE_CNT.
Also there are 2 places where "3" is used rather
than the macro - fix them.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lnet/libcfs/tracefile.c | 15 +++++++--------
drivers/staging/lustre/lnet/libcfs/tracefile.h | 4 ++--
2 files changed, 9 insertions(+), 10 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index fbd05905e1c4..cdef67391a72 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -119,7 +119,7 @@ static void put_pages_on_tcd_daemon_list(struct page_collection *pc,
int cfs_trace_lock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
__acquires(&tcd->tc_lock)
{
- __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_MAX);
+ __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_CNT);
if (tcd->tcd_type == CFS_TCD_TYPE_IRQ)
spin_lock_irqsave(&tcd->tcd_lock, tcd->tcd_lock_flags);
else if (tcd->tcd_type == CFS_TCD_TYPE_SOFTIRQ)
@@ -134,7 +134,7 @@ int cfs_trace_lock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
__releases(&tcd->tcd_lock)
{
- __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_MAX);
+ __LASSERT(tcd->tcd_type < CFS_TCD_TYPE_CNT);
if (tcd->tcd_type == CFS_TCD_TYPE_IRQ)
spin_unlock_irqrestore(&tcd->tcd_lock, tcd->tcd_lock_flags);
else if (tcd->tcd_type == CFS_TCD_TYPE_SOFTIRQ)
@@ -169,14 +169,13 @@ static inline void cfs_trace_put_tcd(struct cfs_trace_cpu_data *tcd)
}
/* percents to share the total debug memory for each type */
-static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
+static unsigned int pages_factor[CFS_TCD_TYPE_CNT] = {
80, /* 80% pages for CFS_TCD_TYPE_PROC */
10, /* 10% pages for CFS_TCD_TYPE_SOFTIRQ */
10 /* 10% pages for CFS_TCD_TYPE_IRQ */
};
-char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
-
+char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_CNT];
enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
{
@@ -1294,7 +1293,7 @@ int cfs_tracefile_init(int max_pages)
/* initialize trace_data */
memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
- for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
+ for (i = 0; i < CFS_TCD_TYPE_CNT; i++) {
cfs_trace_data[i] =
kmalloc_array(num_possible_cpus(),
sizeof(union cfs_trace_data_union),
@@ -1323,7 +1322,7 @@ int cfs_tracefile_init(int max_pages)
}
for (i = 0; i < num_possible_cpus(); i++)
- for (j = 0; j < 3; j++) {
+ for (j = 0; j < CFS_TCD_TYPE_CNT; j++) {
cfs_trace_console_buffers[i][j] =
kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
GFP_KERNEL);
@@ -1377,7 +1376,7 @@ static void cfs_trace_cleanup(void)
trace_cleanup_on_all_cpus();
for (i = 0; i < num_possible_cpus(); i++)
- for (j = 0; j < 3; j++) {
+ for (j = 0; j < CFS_TCD_TYPE_CNT; j++) {
kfree(cfs_trace_console_buffers[i][j]);
cfs_trace_console_buffers[i][j] = NULL;
}
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
index f49a9ba00864..23faecf886c1 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
@@ -46,7 +46,7 @@ enum cfs_trace_buf_type {
CFS_TCD_TYPE_PROC = 0,
CFS_TCD_TYPE_SOFTIRQ,
CFS_TCD_TYPE_IRQ,
- CFS_TCD_TYPE_MAX
+ CFS_TCD_TYPE_CNT
};
#define TRACEFILE_NAME_SIZE 1024
@@ -200,7 +200,7 @@ void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
const char *buf, int len, const char *file,
const char *fn);
-extern char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
+extern char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_CNT];
enum cfs_trace_buf_type cfs_trace_buf_idx_get(void);
static inline char *
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (8 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 04/24] kbuild: disable KBUILD_MODNAME when building for mod.a NeilBrown
@ 2018-06-15 7:11 ` NeilBrown
2018-06-21 1:38 ` James Simmons
2018-06-15 7:11 ` [lustre-devel] [PATCH 13/24] lustre: discard NO_QUOTA and QUOTA_OK NeilBrown
` (14 subsequent siblings)
24 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-15 7:11 UTC (permalink / raw)
To: lustre-devel
As well as CFS_TCD_TYPE_CNT we have TCD_MAX_TYPES which has a larger
value but a similar meaning. Discard it and just use
CFS_TCD_TYPE_CNT.
Two places relied on the fact that TCD_MAX_TYPES was larger and so
there would be NULLs at the end of the array. Change
them to check the array size properly.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lnet/libcfs/tracefile.c | 8 ++++----
drivers/staging/lustre/lnet/libcfs/tracefile.h | 3 +--
2 files changed, 5 insertions(+), 6 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index cdef67391a72..555ade0ddafb 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -50,7 +50,7 @@
#include "tracefile.h"
/* XXX move things up to the top, comment */
-union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS] __cacheline_aligned;
+union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS] __cacheline_aligned;
char cfs_tracefile[TRACEFILE_NAME_SIZE];
long long cfs_tracefile_size = CFS_TRACEFILE_SIZE;
@@ -145,8 +145,8 @@ void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
spin_unlock(&tcd->tcd_lock);
}
-#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
- for (i = 0; cfs_trace_data[i] && \
+#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
+ for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i] && \
(tcd = &(*cfs_trace_data[i])[cpu].tcd) && \
cfs_trace_lock_tcd(tcd, 1); cfs_trace_unlock_tcd(tcd, 1), i++)
@@ -1381,7 +1381,7 @@ static void cfs_trace_cleanup(void)
cfs_trace_console_buffers[i][j] = NULL;
}
- for (i = 0; cfs_trace_data[i]; i++) {
+ for (i = 0; cfs_trace_data[i] && i < CFS_TCD_TYPE_CNT; i++) {
kfree(cfs_trace_data[i]);
cfs_trace_data[i] = NULL;
}
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
index 23faecf886c1..a09d5b9349f0 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
@@ -184,8 +184,7 @@ union cfs_trace_data_union {
char __pad[L1_CACHE_ALIGN(sizeof(struct cfs_trace_cpu_data))];
};
-#define TCD_MAX_TYPES 8
-extern union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS];
+extern union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS];
#define cfs_tcd_for_each(tcd, i, j) \
for (i = 0; cfs_trace_data[i]; i++) \
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
2018-06-15 7:11 ` [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page NeilBrown
@ 2018-06-16 0:17 ` James Simmons
2018-06-16 0:52 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-16 0:17 UTC (permalink / raw)
To: lustre-devel
> Carrying a local truncate_complete_page() is not good
> for maintainability. Linux now used truncate_inode_page,
> so use that instead.
> For correct use, we need to pass the mapping to mdc_release_page()
> rather than trusting page->mapping.
Nak: truncate_inode_page is not exported so it fails to build when lustre
is configured as a module.
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
> .../lustre/include/lustre_patchless_compat.h | 16 ----------------
> drivers/staging/lustre/lustre/llite/dir.c | 3 +--
> .../staging/lustre/lustre/llite/llite_internal.h | 4 ++--
> drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 ++++++------
> 4 files changed, 9 insertions(+), 26 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> index 298476ea7557..dca1e3dbd183 100644
> --- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> +++ b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> @@ -41,22 +41,6 @@
> #include <linux/hash.h>
> #include <linux/pagemap.h>
>
> -#define ll_delete_from_page_cache(page) delete_from_page_cache(page)
> -
> -static inline void
> -truncate_complete_page(struct address_space *mapping, struct page *page)
> -{
> - if (page->mapping != mapping)
> - return;
> -
> - if (PagePrivate(page))
> - page->mapping->a_ops->invalidatepage(page, 0, PAGE_SIZE);
> -
> - cancel_dirty_page(page);
> - ClearPageMappedToDisk(page);
> - ll_delete_from_page_cache(page);
> -}
> -
> #ifndef ATTR_CTIME_SET
> /*
> * set ATTR_CTIME_SET to a high value to avoid any risk of collision with other
> diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
> index 688dddf3ca47..cc5869cf9f93 100644
> --- a/drivers/staging/lustre/lustre/llite/dir.c
> +++ b/drivers/staging/lustre/lustre/llite/dir.c
> @@ -167,8 +167,7 @@ void ll_release_page(struct inode *inode, struct page *page, bool remove)
>
> if (remove) {
> lock_page(page);
> - if (likely(page->mapping))
> - truncate_complete_page(page->mapping, page);
> + truncate_inode_page(inode->i_mapping, page);
> unlock_page(page);
> }
> put_page(page);
> diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
> index c08a6e14b6d7..28cff58690d6 100644
> --- a/drivers/staging/lustre/lustre/llite/llite_internal.h
> +++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
> @@ -938,11 +938,11 @@ static inline void ll_invalidate_page(struct page *vmpage)
> return;
>
> /*
> - * truncate_complete_page() calls
> + * truncate_inode_page() calls
> * a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
> */
> ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
> - truncate_complete_page(mapping, vmpage);
> + truncate_inode_page(mapping, vmpage);
> }
>
> #define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
> diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
> index cff31cb0a9ac..856e6a376852 100644
> --- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
> +++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
> @@ -922,12 +922,12 @@ static int mdc_getpage(struct obd_export *exp, const struct lu_fid *fid,
> return 0;
> }
>
> -static void mdc_release_page(struct page *page, int remove)
> +static void mdc_release_page(struct address_space *mapping,
> + struct page *page, int remove)
> {
> if (remove) {
> lock_page(page);
> - if (likely(page->mapping))
> - truncate_complete_page(page->mapping, page);
> + truncate_inode_page(mapping, page);
> unlock_page(page);
> }
> put_page(page);
> @@ -981,7 +981,7 @@ static struct page *mdc_page_locate(struct address_space *mapping, __u64 *hash,
> offset, *start, *end, *hash);
> if (*hash > *end) {
> kunmap(page);
> - mdc_release_page(page, 0);
> + mdc_release_page(mapping, page, 0);
> page = NULL;
> } else if (*end != *start && *hash == *end) {
> /*
> @@ -991,7 +991,7 @@ static struct page *mdc_page_locate(struct address_space *mapping, __u64 *hash,
> * fetch the page we want.
> */
> kunmap(page);
> - mdc_release_page(page,
> + mdc_release_page(mapping, page,
> le32_to_cpu(dp->ldp_flags) & LDF_COLLIDE);
> page = NULL;
> }
> @@ -1365,7 +1365,7 @@ static int mdc_read_page(struct obd_export *exp, struct md_op_data *op_data,
> return rc;
> fail:
> kunmap(page);
> - mdc_release_page(page, 1);
> + mdc_release_page(mapping, page, 1);
> rc = -EIO;
> goto out_unlock;
> }
>
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
2018-06-16 0:17 ` James Simmons
@ 2018-06-16 0:52 ` NeilBrown
2018-06-16 0:59 ` Oleg Drokin
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-16 0:52 UTC (permalink / raw)
To: lustre-devel
On Sat, Jun 16 2018, James Simmons wrote:
>> Carrying a local truncate_complete_page() is not good
>> for maintainability. Linux now used truncate_inode_page,
>> so use that instead.
>> For correct use, we need to pass the mapping to mdc_release_page()
>> rather than trusting page->mapping.
>
> Nak: truncate_inode_page is not exported so it fails to build when lustre
> is configured as a module.
Bother, nor it is.
We could probably use truncate_inode_pages_range() but it isn't a
perfect fit, which seems to suggest that lustre is trying to do
something that no other filesystem finds the need to do.
I'll dig deeper and work out what is really going on.
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180616/0991e7c2/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
2018-06-16 0:52 ` NeilBrown
@ 2018-06-16 0:59 ` Oleg Drokin
2018-06-18 2:03 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: Oleg Drokin @ 2018-06-16 0:59 UTC (permalink / raw)
To: lustre-devel
> On Jun 15, 2018, at 8:52 PM, NeilBrown <neilb@suse.com> wrote:
>
> On Sat, Jun 16 2018, James Simmons wrote:
>
>>> Carrying a local truncate_complete_page() is not good
>>> for maintainability. Linux now used truncate_inode_page,
>>> so use that instead.
>>> For correct use, we need to pass the mapping to mdc_release_page()
>>> rather than trusting page->mapping.
>>
>> Nak: truncate_inode_page is not exported so it fails to build when lustre
>> is configured as a module.
>
> Bother, nor it is.
> We could probably use truncate_inode_pages_range() but it isn't a
> perfect fit, which seems to suggest that lustre is trying to do
> something that no other filesystem finds the need to do.
> I'll dig deeper and work out what is really going on.
Lustre evicts pages from mapping when the lock that covers the pages
is canceled.
Normally this only applies to data pages, and I see this is in mdc which is for metadata.
For metadata the only pages we hold are readdir pages and the lock currently
covers entire dir so lock cancelation should not need this.
So looking at the code the other two users are readdir where we failed to populate page
somehow and now need to throw it out and then a collision event for hashes, it appears.
Both of these should apply to other filesystems I am sure.
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
2018-06-16 0:59 ` Oleg Drokin
@ 2018-06-18 2:03 ` NeilBrown
2018-06-25 0:02 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-18 2:03 UTC (permalink / raw)
To: lustre-devel
On Fri, Jun 15 2018, Oleg Drokin wrote:
>> On Jun 15, 2018, at 8:52 PM, NeilBrown <neilb@suse.com> wrote:
>>
>> On Sat, Jun 16 2018, James Simmons wrote:
>>
>>>> Carrying a local truncate_complete_page() is not good
>>>> for maintainability. Linux now used truncate_inode_page,
>>>> so use that instead.
>>>> For correct use, we need to pass the mapping to mdc_release_page()
>>>> rather than trusting page->mapping.
>>>
>>> Nak: truncate_inode_page is not exported so it fails to build when lustre
>>> is configured as a module.
>>
>> Bother, nor it is.
>> We could probably use truncate_inode_pages_range() but it isn't a
>> perfect fit, which seems to suggest that lustre is trying to do
>> something that no other filesystem finds the need to do.
>> I'll dig deeper and work out what is really going on.
>
> Lustre evicts pages from mapping when the lock that covers the pages
> is canceled.
> Normally this only applies to data pages, and I see this is in mdc which is for metadata.
The invalidation of data pages is presumably where vvp_page_discard()
calls ll_invalidate_page() [which is not the same as ll_invalidatepage() :-( ]
ll_invalidate_page does almost exactly the same as
truncate_inode_page(), which isn't exported.
We cannot just call
truncate_inode_pages_range(page->mapping,
page->index<<PAGE_SHIFT, page->index<<PAGE_SHIFT + 1);
as that will try to lock the page, which is already locked.
We could use generic_error_remove_page() instead (that is exported).
This condition isn't exactly an error, but it is a situation where
memory soon will not contain correct data.
afs uses this in afs_kill_pages() though the purpose is different.
cepfs uses it in writepages_finish in a context described as:
/*
* We lost the cache cap, need to truncate the page before
* it is unlocked, otherwise we'd truncate it later in the
* page truncation thread, possibly losing some data that
* raced its way in
*/
which sounds similar to what lustre is doing.
NFS can only lose a lock (delegation) on the whole file, not for
individual pages, so it just invalidates a range.
I think I'd like to change lustre to use generic_error_remove_page
instead of ll_invalidate_page() - for S_ISREG mappings.
>
> For metadata the only pages we hold are readdir pages and the lock currently
> covers entire dir so lock cancelation should not need this.
>
> So looking at the code the other two users are readdir where we failed to populate page
> somehow and now need to throw it out and then a collision event for hashes, it appears.
> Both of these should apply to other filesystems I am sure.
In the directory case there is no invalidatepage operation, so the
current truncate_complete_page() function just does:
cancel_dirty_page(page);
ClearPageMappedToDisk(page);
ll_delete_from_page_cache(page);
I'd guess that lustre directory pages don't get PageMappedToDisk set,
and in this situation there aren't dirty.
So this just boils down to
delete_from_page_cache(page)
Do you know if my guesses are correct? If they are, lets just use
delete_from_page_cache() for unwanted directory pages.
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180618/b06cae4e/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid
2018-06-15 7:11 ` [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid NeilBrown
@ 2018-06-21 1:35 ` James Simmons
2018-06-21 4:19 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-21 1:35 UTC (permalink / raw)
To: lustre-devel
> Using spare bits in iattr.ia_valid is not safe as the
> bits might get used for some other purpose without
> us noticing.
> lustre currently used 4 spare bit, but they are all
> defined in different places so this isn't immediately
> obvious.
>
> This patch brings all those bit definitions together
> and defined a new op_data field (op_xvalid) to hold
> the extra validity bits.
>
> It also replaces sa_valid in struct cl_setattr_io
> with sa_avalid and sa_xvalid. Changing the name is
> helpful as sa_valid already has another use within
> lustre.
>
> Signed-off-by: NeilBrown <neilb@suse.com>
Nak: I see regressions with this patch.
[12368.453655] Lustre: DEBUG MARKER: == sanity test 240: race between ldlm
enqueue and the connection RPC (no ASSERT) ===================== 21:
16:30 (1529543790)
[12368.760832] BUG: Dentry
000000002646a847{i=200004282000008,n=f237.sanity} still in use (1)
[unmount of lustre lustre]
[12368.773746] WARNING: CPU: 1 PID: 10861 at fs/dcache.c:1514
umount_check+0x66/0x70
[12368.783364] Modules linked in: ext4(E) mbcache(E) jbd2(E) loop(E)
lustre(CE) obdecho(CE) mgc(CE) lov(CE) osc(CE) mdc(CE) lmv(CE) fid(CE) fld
(CE) ptlrpc(CE) obdclass(CE) ksocklnd(CE) sha512_generic(E) lnet(CE)
libcfs(CE) xt_comment(E) xt_multiport(E) nfsv3(E) nfs_acl(E) sg(E) ip6t_RE
JECT(E) nf_reject_ipv6(E) nf_conntrack_ipv6(E) nf_defrag_ipv6(E)
ip6table_filter(E) ip6_tables(E) ipt_REJECT(E) nf_reject_ipv4(E)
nf_conntrack_
ipv4(E) nf_defrag_ipv4(E) xt_conntrack(E) nf_conntrack(E) libcrc32c(E)
iptable_filter(E) dm_mirror(E) dm_region_hash(E) dm_log(E) dm_mod(E) ib_
isert(E) iscsi_target_mod(E) mgag200(E) ttm(E) target_core_mod(E)
drm_kms_helper(E) rpcrdma(E) x86_pkg_temp_thermal(E) drm(E)
crc32_pclmul(E) i
b_ucm(E) rdma_ucm(E) ib_uverbs(E) fb_sys_fops(E) syscopyarea(E) ib_iser(E)
mpt3sas(E) sysfillrect(E)
[12368.869470] ipmi_si(E) sysimgblt(E) raid_class(E) ib_umad(E)
rdma_cm(E) ipmi_devintf(E) scsi_transport_sas(E) wmi(E) ib_ipoib(E)
ipmi_msgha
ndler(E) iw_cm(E) libiscsi(E) ib_cm(E) scsi_transport_iscsi(E) mlx4_ib(E)
ib_core(E) binfmt_misc(E) ip_tables(E) rpcsec_gss_krb5(E) auth_rpcgss
(E) nfsv4(E) dns_resolver(E) nfs(E) lockd(E) grace(E) fscache(E)
mlx4_en(E) mlx4_core(E) igb(E) i2c_algo_bit(E) devlink(E) crc32c_intel(E)
i2c_
core(E) sunrpc(E)
[12368.917172] CPU: 1 PID: 10861 Comm: umount Tainted: G WC E
4.17.0+ #1
[12368.926825] Hardware name: Supermicro X9DRT/X9DRT, BIOS 3.0a 02/19/2014
[12368.935595] RIP: 0010:umount_check+0x66/0x70
[12368.941993] RSP: 0018:ffffc90007313da8 EFLAGS: 00010286
[12368.949342] RAX: 000000000000006a RBX: ffff8810557dcb40 RCX:
0000000000000000
[12368.958619] RDX: 0000000000000001 RSI: ffff88085fc954f8 RDI:
ffff88085fc954f8
[12368.967868] RBP: ffff880fec9ef600 R08: 0000000000000000 R09:
0000000000000896
[12368.977102] R10: 00000000000003ff R11: 0000000000aaaaaa R12:
ffffffff8124a320
[12368.986315] R13: ffff8810557dcbe0 R14: ffff880fec9ef690 R15:
ffff8810557dcbe0
[12368.995505] FS: 00007fc174af5880(0000) GS:ffff88085fc80000(0000)
knlGS:0000000000000000
[12369.005631] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[12369.013395] CR2: 00007fc174b04000 CR3: 0000000822922002 CR4:
00000000000606e0
[12369.022530] Call Trace:
[12369.026962] d_walk+0xe1/0x270
[12369.031970] do_one_tree+0x22/0x40
[12369.037280] shrink_dcache_for_umount+0x28/0x80
[12369.043685] generic_shutdown_super+0x1a/0x110
[12369.049977] kill_anon_super+0xe/0x20
[12369.055463] deactivate_locked_super+0x39/0x70
[12369.061708] cleanup_mnt+0x3b/0x70
[12369.066871] task_work_run+0x7c/0xa0
[12369.072174] exit_to_usermode_loop+0x5a/0x7e
[12369.078140] do_syscall_64+0x2b7/0x347
[12369.083547] ? do_page_fault+0x2d/0x110
[12369.089008] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[12369.095655] RIP: 0033:0x7fc173d69bd7
[12369.100789] RSP: 002b:00007ffc4c89fcc8 EFLAGS: 00000246 ORIG_RAX:
00000000000000a6
[12369.109902] RAX: 0000000000000000 RBX: 000056521cd63040 RCX:
00007fc173d69bd7
[12369.118564] RDX: 0000000000000001 RSI: 0000000000000000 RDI:
000056521cd642b0
[12369.127188] RBP: 000056521cd642b0 R08: 0000000000000000 R09:
000000000000000f
[12369.135778] R10: 00007ffc4c89f720 R11: 0000000000000246 R12:
00007fc1748e4d58
[12369.144336] R13: 0000000000000000 R14: 000056521cd648a0 R15:
000056521cd63040
[12369.152868] Code: 00 00 48 8b 40 28 4c 8b 08 48 8b 46 30 48 85 c0 74 1d
48 8b 50 40 48 89 0c 24 48 c7 c7 58 5c e6 81 48 89 f1 31 c0 e8 38 cf ea ff
<0f> 0b eb b2 31 d2 eb e3 66 90 66 66 66 66 90 55 48 89 fd 53 48
[12369.174646] ---[ end trace 053c2a5a718d3440 ]---
[12369.385564] Lustre: Unmounted lustre-client
[12369.393247] VFS: Busy inodes after unmount of lustre. Self-destruct in
5 seconds. Have a nice day...
When I remove this patch things go back to normal. This will not show up
if you do a ONLY="240" sh ./sanity.sh. You have to run the sanity.sh in
total to make this show up.
> ---
> drivers/staging/lustre/lustre/include/cl_object.h | 5 +++-
> .../staging/lustre/lustre/include/lustre_compat.h | 6 -----
> .../lustre/include/lustre_patchless_compat.h | 8 -------
> drivers/staging/lustre/lustre/include/obd.h | 7 ++++++
> drivers/staging/lustre/lustre/llite/file.c | 12 ++++++----
> drivers/staging/lustre/lustre/llite/lcommon_cl.c | 5 +++-
> .../staging/lustre/lustre/llite/llite_internal.h | 5 +++-
> drivers/staging/lustre/lustre/llite/llite_lib.c | 21 ++++++++++--------
> drivers/staging/lustre/lustre/llite/vvp_io.c | 8 +++----
> drivers/staging/lustre/lustre/lmv/lmv_obd.c | 5 +++-
> drivers/staging/lustre/lustre/lov/lov_io.c | 3 ++-
> drivers/staging/lustre/lustre/mdc/mdc_lib.c | 13 ++++++-----
> drivers/staging/lustre/lustre/osc/osc_io.c | 23 ++++++++++----------
> 13 files changed, 63 insertions(+), 58 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/include/cl_object.h b/drivers/staging/lustre/lustre/include/cl_object.h
> index 6f7b991be809..caa66763c70d 100644
> --- a/drivers/staging/lustre/lustre/include/cl_object.h
> +++ b/drivers/staging/lustre/lustre/include/cl_object.h
> @@ -1773,7 +1773,8 @@ struct cl_io {
> struct cl_setattr_io {
> struct ost_lvb sa_attr;
> unsigned int sa_attr_flags;
> - unsigned int sa_valid;
> + unsigned int sa_avalid;
> + unsigned int sa_xvalid;
> int sa_stripe_index;
> const struct lu_fid *sa_parent_fid;
> } ci_setattr;
> @@ -2304,7 +2305,7 @@ static inline int cl_io_is_mkwrite(const struct cl_io *io)
> static inline int cl_io_is_trunc(const struct cl_io *io)
> {
> return io->ci_type == CIT_SETATTR &&
> - (io->u.ci_setattr.sa_valid & ATTR_SIZE);
> + (io->u.ci_setattr.sa_avalid & ATTR_SIZE);
> }
>
> struct cl_io *cl_io_top(struct cl_io *io);
> diff --git a/drivers/staging/lustre/lustre/include/lustre_compat.h b/drivers/staging/lustre/lustre/include/lustre_compat.h
> index 3c6db0d632dc..6d1a433e32c5 100644
> --- a/drivers/staging/lustre/lustre/include/lustre_compat.h
> +++ b/drivers/staging/lustre/lustre/include/lustre_compat.h
> @@ -41,12 +41,6 @@
>
> #include <lustre_patchless_compat.h>
>
> -/*
> - * set ATTR_BLOCKS to a high value to avoid any risk of collision with other
> - * ATTR_* attributes (see bug 13828)
> - */
> -#define ATTR_BLOCKS (1 << 27)
> -
> #define current_ngroups current_cred()->group_info->ngroups
> #define current_groups current_cred()->group_info->small_block
>
> diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> index dca1e3dbd183..1bbe1a779ae9 100644
> --- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> +++ b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> @@ -41,12 +41,4 @@
> #include <linux/hash.h>
> #include <linux/pagemap.h>
>
> -#ifndef ATTR_CTIME_SET
> -/*
> - * set ATTR_CTIME_SET to a high value to avoid any risk of collision with other
> - * ATTR_* attributes (see bug 13828)
> - */
> -#define ATTR_CTIME_SET (1 << 28)
> -#endif
> -
> #endif /* LUSTRE_PATCHLESS_COMPAT_H */
> diff --git a/drivers/staging/lustre/lustre/include/obd.h b/drivers/staging/lustre/lustre/include/obd.h
> index b1907bbffb19..d38b6bc6b246 100644
> --- a/drivers/staging/lustre/lustre/include/obd.h
> +++ b/drivers/staging/lustre/lustre/include/obd.h
> @@ -723,6 +723,7 @@ struct md_op_data {
>
> /* iattr fields and blocks. */
> struct iattr op_attr;
> + unsigned int op_xvalid; /* eXtra validity flags */
> unsigned int op_attr_flags;
> __u64 op_valid;
> loff_t op_attr_blocks;
> @@ -751,6 +752,12 @@ struct md_op_data {
> __u32 op_default_stripe_offset;
> };
>
> +/* Flags for op_xvalid */
> +#define OP_ATTR_CTIME_SET (1 << 0)
> +#define OP_ATTR_BLOCKS (1 << 1)
> +#define OP_ATTR_OWNEROVERRIDE (1 << 2)
> +#define OP_ATTR_FLAGS (1 << 3)
> +
> struct md_callback {
> int (*md_blocking_ast)(struct ldlm_lock *lock,
> struct ldlm_lock_desc *desc,
> diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
> index e888ed6e74bc..e642996db53f 100644
> --- a/drivers/staging/lustre/lustre/llite/file.c
> +++ b/drivers/staging/lustre/lustre/llite/file.c
> @@ -93,9 +93,10 @@ static void ll_prepare_close(struct inode *inode, struct md_op_data *op_data,
> op_data->op_attr.ia_mtime = inode->i_mtime;
> op_data->op_attr.ia_ctime = inode->i_ctime;
> op_data->op_attr.ia_size = i_size_read(inode);
> - op_data->op_attr.ia_valid |= ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET |
> - ATTR_MTIME | ATTR_MTIME_SET |
> - ATTR_CTIME | ATTR_CTIME_SET;
> + op_data->op_attr.ia_valid |= (ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET |
> + ATTR_MTIME | ATTR_MTIME_SET |
> + ATTR_CTIME);
> + op_data->op_xvalid |= OP_ATTR_CTIME_SET;
> op_data->op_attr_blocks = inode->i_blocks;
> op_data->op_attr_flags = ll_inode_to_ext_flags(inode->i_flags);
> op_data->op_handle = och->och_fh;
> @@ -161,7 +162,8 @@ static int ll_close_inode_openhandle(struct inode *inode,
> op_data->op_bias |= MDS_HSM_RELEASE;
> op_data->op_data_version = *(__u64 *)data;
> op_data->op_lease_handle = och->och_lease_handle;
> - op_data->op_attr.ia_valid |= ATTR_SIZE | ATTR_BLOCKS;
> + op_data->op_attr.ia_valid |= ATTR_SIZE;
> + op_data->op_xvalid |= OP_ATTR_BLOCKS;
> break;
>
> default:
> @@ -1906,7 +1908,7 @@ static int ll_hsm_import(struct inode *inode, struct file *file,
>
> inode_lock(inode);
>
> - rc = ll_setattr_raw(file->f_path.dentry, attr, true);
> + rc = ll_setattr_raw(file->f_path.dentry, attr, 0, true);
> if (rc == -ENODATA)
> rc = 0;
>
> diff --git a/drivers/staging/lustre/lustre/llite/lcommon_cl.c b/drivers/staging/lustre/lustre/llite/lcommon_cl.c
> index d7ea39ce0cb2..9d1f3881f860 100644
> --- a/drivers/staging/lustre/lustre/llite/lcommon_cl.c
> +++ b/drivers/staging/lustre/lustre/llite/lcommon_cl.c
> @@ -80,7 +80,7 @@ u16 cl_inode_fini_refcheck;
> static DEFINE_MUTEX(cl_inode_fini_guard);
>
> int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
> - unsigned int attr_flags)
> + unsigned int xvalid, unsigned int attr_flags)
> {
> struct lu_env *env;
> struct cl_io *io;
> @@ -100,7 +100,8 @@ int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
> io->u.ci_setattr.sa_attr.lvb_ctime = LTIME_S(attr->ia_ctime);
> io->u.ci_setattr.sa_attr.lvb_size = attr->ia_size;
> io->u.ci_setattr.sa_attr_flags = attr_flags;
> - io->u.ci_setattr.sa_valid = attr->ia_valid;
> + io->u.ci_setattr.sa_avalid = attr->ia_valid;
> + io->u.ci_setattr.sa_xvalid = xvalid;
> io->u.ci_setattr.sa_parent_fid = lu_object_fid(&obj->co_lu);
>
> again:
> diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
> index 28cff58690d6..b750a48fd0e1 100644
> --- a/drivers/staging/lustre/lustre/llite/llite_internal.h
> +++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
> @@ -805,7 +805,8 @@ void ll_kill_super(struct super_block *sb);
> struct inode *ll_inode_from_resource_lock(struct ldlm_lock *lock);
> void ll_dir_clear_lsm_md(struct inode *inode);
> void ll_clear_inode(struct inode *inode);
> -int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import);
> +int ll_setattr_raw(struct dentry *dentry, struct iattr *attr,
> + unsigned int xvalid, bool hsm_import);
> int ll_setattr(struct dentry *de, struct iattr *attr);
> int ll_statfs(struct dentry *de, struct kstatfs *sfs);
> int ll_statfs_internal(struct super_block *sb, struct obd_statfs *osfs,
> @@ -1330,7 +1331,7 @@ int ll_getparent(struct file *file, struct getparent __user *arg);
>
> /* lcommon_cl.c */
> int cl_setattr_ost(struct cl_object *obj, const struct iattr *attr,
> - unsigned int attr_flags);
> + unsigned int xvalid, unsigned int attr_flags);
>
> extern struct lu_env *cl_inode_fini_env;
> extern u16 cl_inode_fini_refcheck;
> diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
> index 36066c839160..cd0f2517a3b3 100644
> --- a/drivers/staging/lustre/lustre/llite/llite_lib.c
> +++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
> @@ -1429,7 +1429,8 @@ static int ll_md_setattr(struct dentry *dentry, struct md_op_data *op_data)
> *
> * In case of HSMimport, we only set attr on MDS.
> */
> -int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
> +int ll_setattr_raw(struct dentry *dentry, struct iattr *attr,
> + unsigned int xvalid, bool hsm_import)
> {
> struct inode *inode = d_inode(dentry);
> struct ll_inode_info *lli = ll_i2info(inode);
> @@ -1470,7 +1471,7 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
> /* We mark all of the fields "set" so MDS/OST does not re-set them */
> if (attr->ia_valid & ATTR_CTIME) {
> attr->ia_ctime = current_time(inode);
> - attr->ia_valid |= ATTR_CTIME_SET;
> + xvalid |= OP_ATTR_CTIME_SET;
> }
> if (!(attr->ia_valid & ATTR_ATIME_SET) &&
> (attr->ia_valid & ATTR_ATIME)) {
> @@ -1506,12 +1507,13 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
> * If we are changing file size, file content is
> * modified, flag it.
> */
> - attr->ia_valid |= MDS_OPEN_OWNEROVERRIDE;
> + xvalid |= OP_ATTR_OWNEROVERRIDE;
> op_data->op_bias |= MDS_DATA_MODIFIED;
> clear_bit(LLIF_DATA_MODIFIED, &lli->lli_flags);
> }
>
> op_data->op_attr = *attr;
> + op_data->op_xvalid = xvalid;
>
> rc = ll_md_setattr(dentry, op_data);
> if (rc)
> @@ -1532,7 +1534,8 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
> * setting times to past, but it is necessary due to possible
> * time de-synchronization between MDT inode and OST objects
> */
> - rc = cl_setattr_ost(ll_i2info(inode)->lli_clob, attr, 0);
> + rc = cl_setattr_ost(ll_i2info(inode)->lli_clob,
> + attr, xvalid, 0);
> }
>
> /*
> @@ -1589,10 +1592,11 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr, bool hsm_import)
> int ll_setattr(struct dentry *de, struct iattr *attr)
> {
> int mode = d_inode(de)->i_mode;
> + unsigned int xvalid = 0;
>
> if ((attr->ia_valid & (ATTR_CTIME | ATTR_SIZE | ATTR_MODE)) ==
> (ATTR_CTIME | ATTR_SIZE | ATTR_MODE))
> - attr->ia_valid |= MDS_OPEN_OWNEROVERRIDE;
> + xvalid |= OP_ATTR_OWNEROVERRIDE;
>
> if (((attr->ia_valid & (ATTR_MODE | ATTR_FORCE | ATTR_SIZE)) ==
> (ATTR_SIZE | ATTR_MODE)) &&
> @@ -1613,7 +1617,7 @@ int ll_setattr(struct dentry *de, struct iattr *attr)
> !(attr->ia_valid & ATTR_KILL_SGID))
> attr->ia_valid |= ATTR_KILL_SGID;
>
> - return ll_setattr_raw(de, attr, false);
> + return ll_setattr_raw(de, attr, xvalid, false);
> }
>
> int ll_statfs_internal(struct super_block *sb, struct obd_statfs *osfs,
> @@ -1946,7 +1950,7 @@ int ll_iocontrol(struct inode *inode, struct file *file,
> return PTR_ERR(op_data);
>
> op_data->op_attr_flags = flags;
> - op_data->op_attr.ia_valid |= ATTR_ATTR_FLAG;
> + op_data->op_xvalid |= OP_ATTR_FLAGS;
> rc = md_setattr(sbi->ll_md_exp, op_data, NULL, 0, &req);
> ll_finish_md_op_data(op_data);
> ptlrpc_req_finished(req);
> @@ -1963,8 +1967,7 @@ int ll_iocontrol(struct inode *inode, struct file *file,
> if (!attr)
> return -ENOMEM;
>
> - attr->ia_valid = ATTR_ATTR_FLAG;
> - rc = cl_setattr_ost(obj, attr, flags);
> + rc = cl_setattr_ost(obj, attr, OP_ATTR_FLAGS, flags);
> kfree(attr);
> return rc;
> }
> diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
> index e7a4778e02e4..e65523f7628d 100644
> --- a/drivers/staging/lustre/lustre/llite/vvp_io.c
> +++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
> @@ -556,7 +556,7 @@ static int vvp_io_setattr_lock(const struct lu_env *env,
> if (new_size == 0)
> enqflags = CEF_DISCARD_DATA;
> } else {
> - unsigned int valid = io->u.ci_setattr.sa_valid;
> + unsigned int valid = io->u.ci_setattr.sa_avalid;
>
> if (!(valid & TIMES_SET_FLAGS))
> return 0;
> @@ -603,11 +603,11 @@ static int vvp_io_setattr_time(const struct lu_env *env,
>
> cl_object_attr_lock(obj);
> attr->cat_ctime = io->u.ci_setattr.sa_attr.lvb_ctime;
> - if (io->u.ci_setattr.sa_valid & ATTR_ATIME_SET) {
> + if (io->u.ci_setattr.sa_avalid & ATTR_ATIME_SET) {
> attr->cat_atime = io->u.ci_setattr.sa_attr.lvb_atime;
> valid |= CAT_ATIME;
> }
> - if (io->u.ci_setattr.sa_valid & ATTR_MTIME_SET) {
> + if (io->u.ci_setattr.sa_avalid & ATTR_MTIME_SET) {
> attr->cat_mtime = io->u.ci_setattr.sa_attr.lvb_mtime;
> valid |= CAT_MTIME;
> }
> @@ -632,7 +632,7 @@ static int vvp_io_setattr_start(const struct lu_env *env,
> inode_lock(inode);
> }
>
> - if (io->u.ci_setattr.sa_valid & TIMES_SET_FLAGS)
> + if (io->u.ci_setattr.sa_avalid & TIMES_SET_FLAGS)
> return vvp_io_setattr_time(env, ios);
>
> return 0;
> diff --git a/drivers/staging/lustre/lustre/lmv/lmv_obd.c b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
> index 65f94e6ecaad..9f427a7c7b03 100644
> --- a/drivers/staging/lustre/lustre/lmv/lmv_obd.c
> +++ b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
> @@ -1983,8 +1983,9 @@ static int lmv_setattr(struct obd_export *exp, struct md_op_data *op_data,
> struct lmv_obd *lmv = &obd->u.lmv;
> struct lmv_tgt_desc *tgt;
>
> - CDEBUG(D_INODE, "SETATTR for " DFID ", valid 0x%x\n",
> - PFID(&op_data->op_fid1), op_data->op_attr.ia_valid);
> + CDEBUG(D_INODE, "SETATTR for " DFID ", valid 0x%x/0x%x\n",
> + PFID(&op_data->op_fid1), op_data->op_attr.ia_valid,
> + op_data->op_xvalid);
>
> op_data->op_flags |= MF_MDC_CANCEL_FID1;
> tgt = lmv_find_target(lmv, &op_data->op_fid1);
> diff --git a/drivers/staging/lustre/lustre/lov/lov_io.c b/drivers/staging/lustre/lustre/lov/lov_io.c
> index b823f8a21856..50982848bca4 100644
> --- a/drivers/staging/lustre/lustre/lov/lov_io.c
> +++ b/drivers/staging/lustre/lustre/lov/lov_io.c
> @@ -77,7 +77,8 @@ static void lov_io_sub_inherit(struct cl_io *io, struct lov_io *lio,
> io->u.ci_setattr.sa_attr = parent->u.ci_setattr.sa_attr;
> io->u.ci_setattr.sa_attr_flags =
> parent->u.ci_setattr.sa_attr_flags;
> - io->u.ci_setattr.sa_valid = parent->u.ci_setattr.sa_valid;
> + io->u.ci_setattr.sa_avalid = parent->u.ci_setattr.sa_avalid;
> + io->u.ci_setattr.sa_xvalid = parent->u.ci_setattr.sa_xvalid;
> io->u.ci_setattr.sa_stripe_index = stripe;
> io->u.ci_setattr.sa_parent_fid =
> parent->u.ci_setattr.sa_parent_fid;
> diff --git a/drivers/staging/lustre/lustre/mdc/mdc_lib.c b/drivers/staging/lustre/lustre/mdc/mdc_lib.c
> index d582968987ff..ad7c13be5e3c 100644
> --- a/drivers/staging/lustre/lustre/mdc/mdc_lib.c
> +++ b/drivers/staging/lustre/lustre/mdc/mdc_lib.c
> @@ -234,7 +234,7 @@ void mdc_open_pack(struct ptlrpc_request *req, struct md_op_data *op_data,
> set_mrc_cr_flags(rec, cr_flags);
> }
>
> -static inline __u64 attr_pack(unsigned int ia_valid)
> +static inline __u64 attr_pack(unsigned int ia_valid, unsigned int ia_xvalid)
> {
> __u64 sa_valid = 0;
>
> @@ -258,19 +258,19 @@ static inline __u64 attr_pack(unsigned int ia_valid)
> sa_valid |= MDS_ATTR_MTIME_SET;
> if (ia_valid & ATTR_FORCE)
> sa_valid |= MDS_ATTR_FORCE;
> - if (ia_valid & ATTR_ATTR_FLAG)
> + if (ia_xvalid & OP_ATTR_FLAGS)
> sa_valid |= MDS_ATTR_ATTR_FLAG;
> if (ia_valid & ATTR_KILL_SUID)
> sa_valid |= MDS_ATTR_KILL_SUID;
> if (ia_valid & ATTR_KILL_SGID)
> sa_valid |= MDS_ATTR_KILL_SGID;
> - if (ia_valid & ATTR_CTIME_SET)
> + if (ia_xvalid & OP_ATTR_CTIME_SET)
> sa_valid |= MDS_ATTR_CTIME_SET;
> if (ia_valid & ATTR_OPEN)
> sa_valid |= MDS_ATTR_FROM_OPEN;
> - if (ia_valid & ATTR_BLOCKS)
> + if (ia_xvalid & OP_ATTR_BLOCKS)
> sa_valid |= MDS_ATTR_BLOCKS;
> - if (ia_valid & MDS_OPEN_OWNEROVERRIDE)
> + if (ia_xvalid & OP_ATTR_OWNEROVERRIDE)
> /* NFSD hack (see bug 5781) */
> sa_valid |= MDS_OPEN_OWNEROVERRIDE;
> return sa_valid;
> @@ -286,7 +286,8 @@ static void mdc_setattr_pack_rec(struct mdt_rec_setattr *rec,
> rec->sa_suppgid = -1;
>
> rec->sa_fid = op_data->op_fid1;
> - rec->sa_valid = attr_pack(op_data->op_attr.ia_valid);
> + rec->sa_valid = attr_pack(op_data->op_attr.ia_valid,
> + op_data->op_xvalid);
> rec->sa_mode = op_data->op_attr.ia_mode;
> rec->sa_uid = from_kuid(&init_user_ns, op_data->op_attr.ia_uid);
> rec->sa_gid = from_kgid(&init_user_ns, op_data->op_attr.ia_gid);
> diff --git a/drivers/staging/lustre/lustre/osc/osc_io.c b/drivers/staging/lustre/lustre/osc/osc_io.c
> index 67734a8ed331..955525f69958 100644
> --- a/drivers/staging/lustre/lustre/osc/osc_io.c
> +++ b/drivers/staging/lustre/lustre/osc/osc_io.c
> @@ -499,7 +499,8 @@ static int osc_io_setattr_start(const struct lu_env *env,
> struct obdo *oa = &oio->oi_oa;
> struct osc_async_cbargs *cbargs = &oio->oi_cbarg;
> __u64 size = io->u.ci_setattr.sa_attr.lvb_size;
> - unsigned int ia_valid = io->u.ci_setattr.sa_valid;
> + unsigned int ia_avalid = io->u.ci_setattr.sa_avalid;
> + unsigned int ia_xvalid = io->u.ci_setattr.sa_xvalid;
> int result = 0;
>
> /* truncate cache dirty pages first */
> @@ -514,20 +515,20 @@ static int osc_io_setattr_start(const struct lu_env *env,
> struct ost_lvb *lvb = &io->u.ci_setattr.sa_attr;
> unsigned int cl_valid = 0;
>
> - if (ia_valid & ATTR_SIZE) {
> + if (ia_avalid & ATTR_SIZE) {
> attr->cat_size = size;
> attr->cat_kms = size;
> cl_valid = CAT_SIZE | CAT_KMS;
> }
> - if (ia_valid & ATTR_MTIME_SET) {
> + if (ia_avalid & ATTR_MTIME_SET) {
> attr->cat_mtime = lvb->lvb_mtime;
> cl_valid |= CAT_MTIME;
> }
> - if (ia_valid & ATTR_ATIME_SET) {
> + if (ia_avalid & ATTR_ATIME_SET) {
> attr->cat_atime = lvb->lvb_atime;
> cl_valid |= CAT_ATIME;
> }
> - if (ia_valid & ATTR_CTIME_SET) {
> + if (ia_xvalid & OP_ATTR_CTIME_SET) {
> attr->cat_ctime = lvb->lvb_ctime;
> cl_valid |= CAT_CTIME;
> }
> @@ -542,19 +543,19 @@ static int osc_io_setattr_start(const struct lu_env *env,
> obdo_set_parent_fid(oa, io->u.ci_setattr.sa_parent_fid);
> oa->o_stripe_idx = io->u.ci_setattr.sa_stripe_index;
> oa->o_valid |= OBD_MD_FLID | OBD_MD_FLGROUP;
> - if (ia_valid & ATTR_CTIME) {
> + if (ia_avalid & ATTR_CTIME) {
> oa->o_valid |= OBD_MD_FLCTIME;
> oa->o_ctime = attr->cat_ctime;
> }
> - if (ia_valid & ATTR_ATIME) {
> + if (ia_avalid & ATTR_ATIME) {
> oa->o_valid |= OBD_MD_FLATIME;
> oa->o_atime = attr->cat_atime;
> }
> - if (ia_valid & ATTR_MTIME) {
> + if (ia_avalid & ATTR_MTIME) {
> oa->o_valid |= OBD_MD_FLMTIME;
> oa->o_mtime = attr->cat_mtime;
> }
> - if (ia_valid & ATTR_SIZE) {
> + if (ia_avalid & ATTR_SIZE) {
> oa->o_size = size;
> oa->o_blocks = OBD_OBJECT_EOF;
> oa->o_valid |= OBD_MD_FLSIZE | OBD_MD_FLBLOCKS;
> @@ -566,14 +567,14 @@ static int osc_io_setattr_start(const struct lu_env *env,
> } else {
> LASSERT(oio->oi_lockless == 0);
> }
> - if (ia_valid & ATTR_ATTR_FLAG) {
> + if (ia_xvalid & OP_ATTR_FLAGS) {
> oa->o_flags = io->u.ci_setattr.sa_attr_flags;
> oa->o_valid |= OBD_MD_FLFLAGS;
> }
>
> init_completion(&cbargs->opc_sync);
>
> - if (ia_valid & ATTR_SIZE)
> + if (ia_avalid & ATTR_SIZE)
> result = osc_punch_base(osc_export(cl2osc(obj)),
> oa, osc_async_upcall,
> cbargs, PTLRPCD_SET);
>
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c
2018-06-15 7:11 ` [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c NeilBrown
@ 2018-06-21 1:37 ` James Simmons
2018-06-22 3:19 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-21 1:37 UTC (permalink / raw)
To: lustre-devel
> It's good to keep related code together.
>
> Signed-off-by: NeilBrown <neilb@suse.com>
Nak. For some reason this corrupts my dmesg output. Its really strange and
I haven't figured it out yet.
> ---
> drivers/staging/lustre/lnet/libcfs/Makefile | 2
> .../staging/lustre/lnet/libcfs/linux-tracefile.c | 194 --------------------
> drivers/staging/lustre/lnet/libcfs/tracefile.c | 155 ++++++++++++++++
> 3 files changed, 156 insertions(+), 195 deletions(-)
> delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
>
> diff --git a/drivers/staging/lustre/lnet/libcfs/Makefile b/drivers/staging/lustre/lnet/libcfs/Makefile
> index 1b6937c3deee..22c0e0c2e9cf 100644
> --- a/drivers/staging/lustre/lnet/libcfs/Makefile
> +++ b/drivers/staging/lustre/lnet/libcfs/Makefile
> @@ -3,7 +3,7 @@ ccflags-y += -I$(srctree)/drivers/staging/lustre/include
> ccflags-y += -I$(srctree)/drivers/staging/lustre/lustre/include
> ccflags-y += -DKBUILD_MODNAME='"lnet"'
>
> -libcfs-obj-y += linux-tracefile.o linux-debug.o
> +libcfs-obj-y += linux-debug.o
> libcfs-obj-y += linux-crypto.o
> libcfs-obj-y += linux-crypto-adler.o
>
> diff --git a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c b/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
> deleted file mode 100644
> index 3af77228cd7a..000000000000
> --- a/drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
> +++ /dev/null
> @@ -1,194 +0,0 @@
> -// SPDX-License-Identifier: GPL-2.0
> -/*
> - * GPL HEADER START
> - *
> - * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
> - *
> - * This program is free software; you can redistribute it and/or modify
> - * it under the terms of the GNU General Public License version 2 only,
> - * as published by the Free Software Foundation.
> - *
> - * This program is distributed in the hope that it will be useful, but
> - * WITHOUT ANY WARRANTY; without even the implied warranty of
> - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> - * General Public License version 2 for more details (a copy is included
> - * in the LICENSE file that accompanied this code).
> - *
> - * You should have received a copy of the GNU General Public License
> - * version 2 along with this program; If not, see
> - * http://www.gnu.org/licenses/gpl-2.0.html
> - *
> - * GPL HEADER END
> - */
> -/*
> - * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
> - * Use is subject to license terms.
> - *
> - * Copyright (c) 2012, Intel Corporation.
> - */
> -/*
> - * This file is part of Lustre, http://www.lustre.org/
> - * Lustre is a trademark of Sun Microsystems, Inc.
> - */
> -
> -#define DEBUG_SUBSYSTEM S_LNET
> -#define LUSTRE_TRACEFILE_PRIVATE
> -
> -#include <linux/slab.h>
> -#include <linux/mm.h>
> -#include "tracefile.h"
> -
> -/* percents to share the total debug memory for each type */
> -static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
> - 80, /* 80% pages for CFS_TCD_TYPE_PROC */
> - 10, /* 10% pages for CFS_TCD_TYPE_SOFTIRQ */
> - 10 /* 10% pages for CFS_TCD_TYPE_IRQ */
> -};
> -
> -char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
> -
> -int cfs_tracefile_init_arch(void)
> -{
> - int i;
> - int j;
> - struct cfs_trace_cpu_data *tcd;
> -
> - /* initialize trace_data */
> - memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
> - for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
> - cfs_trace_data[i] =
> - kmalloc_array(num_possible_cpus(),
> - sizeof(union cfs_trace_data_union),
> - GFP_KERNEL);
> - if (!cfs_trace_data[i])
> - goto out;
> - }
> -
> - /* arch related info initialized */
> - cfs_tcd_for_each(tcd, i, j) {
> - spin_lock_init(&tcd->tcd_lock);
> - tcd->tcd_pages_factor = pages_factor[i];
> - tcd->tcd_type = i;
> - tcd->tcd_cpu = j;
> - }
> -
> - for (i = 0; i < num_possible_cpus(); i++)
> - for (j = 0; j < 3; j++) {
> - cfs_trace_console_buffers[i][j] =
> - kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
> - GFP_KERNEL);
> -
> - if (!cfs_trace_console_buffers[i][j])
> - goto out;
> - }
> -
> - return 0;
> -
> -out:
> - cfs_tracefile_fini_arch();
> - pr_err("lnet: Not enough memory\n");
> - return -ENOMEM;
> -}
> -
> -void cfs_tracefile_fini_arch(void)
> -{
> - int i;
> - int j;
> -
> - for (i = 0; i < num_possible_cpus(); i++)
> - for (j = 0; j < 3; j++) {
> - kfree(cfs_trace_console_buffers[i][j]);
> - cfs_trace_console_buffers[i][j] = NULL;
> - }
> -
> - for (i = 0; cfs_trace_data[i]; i++) {
> - kfree(cfs_trace_data[i]);
> - cfs_trace_data[i] = NULL;
> - }
> -}
> -
> -enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
> -{
> - if (in_irq())
> - return CFS_TCD_TYPE_IRQ;
> - if (in_softirq())
> - return CFS_TCD_TYPE_SOFTIRQ;
> - return CFS_TCD_TYPE_PROC;
> -}
> -
> -void
> -cfs_set_ptldebug_header(struct ptldebug_header *header,
> - struct libcfs_debug_msg_data *msgdata,
> - unsigned long stack)
> -{
> - struct timespec64 ts;
> -
> - ktime_get_real_ts64(&ts);
> -
> - header->ph_subsys = msgdata->msg_subsys;
> - header->ph_mask = msgdata->msg_mask;
> - header->ph_cpu_id = smp_processor_id();
> - header->ph_type = cfs_trace_buf_idx_get();
> - /* y2038 safe since all user space treats this as unsigned, but
> - * will overflow in 2106
> - */
> - header->ph_sec = (u32)ts.tv_sec;
> - header->ph_usec = ts.tv_nsec / NSEC_PER_USEC;
> - header->ph_stack = stack;
> - header->ph_pid = current->pid;
> - header->ph_line_num = msgdata->msg_line;
> - header->ph_extern_pid = 0;
> -}
> -
> -static char *
> -dbghdr_to_err_string(struct ptldebug_header *hdr)
> -{
> - switch (hdr->ph_subsys) {
> - case S_LND:
> - case S_LNET:
> - return "LNetError";
> - default:
> - return "LustreError";
> - }
> -}
> -
> -static char *
> -dbghdr_to_info_string(struct ptldebug_header *hdr)
> -{
> - switch (hdr->ph_subsys) {
> - case S_LND:
> - case S_LNET:
> - return "LNet";
> - default:
> - return "Lustre";
> - }
> -}
> -
> -void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
> - const char *buf, int len, const char *file,
> - const char *fn)
> -{
> - char *prefix = "Lustre", *ptype = NULL;
> -
> - if (mask & D_EMERG) {
> - prefix = dbghdr_to_err_string(hdr);
> - ptype = KERN_EMERG;
> - } else if (mask & D_ERROR) {
> - prefix = dbghdr_to_err_string(hdr);
> - ptype = KERN_ERR;
> - } else if (mask & D_WARNING) {
> - prefix = dbghdr_to_info_string(hdr);
> - ptype = KERN_WARNING;
> - } else if (mask & (D_CONSOLE | libcfs_printk)) {
> - prefix = dbghdr_to_info_string(hdr);
> - ptype = KERN_INFO;
> - }
> -
> - if (mask & D_CONSOLE) {
> - pr_info("%s%s: %.*s", ptype, prefix, len, buf);
> - } else {
> - pr_info("%s%s: %d:%d:(%s:%d:%s()) %.*s", ptype, prefix,
> - hdr->ph_pid, hdr->ph_extern_pid, file,
> - hdr->ph_line_num, fn, len, buf);
> - }
> -}
> diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
> index 6d567a9ad8ea..617edc291035 100644
> --- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
> +++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
> @@ -168,6 +168,161 @@ static inline void cfs_trace_put_tcd(struct cfs_trace_cpu_data *tcd)
> put_cpu();
> }
>
> +/* percents to share the total debug memory for each type */
> +static unsigned int pages_factor[CFS_TCD_TYPE_MAX] = {
> + 80, /* 80% pages for CFS_TCD_TYPE_PROC */
> + 10, /* 10% pages for CFS_TCD_TYPE_SOFTIRQ */
> + 10 /* 10% pages for CFS_TCD_TYPE_IRQ */
> +};
> +
> +char *cfs_trace_console_buffers[NR_CPUS][CFS_TCD_TYPE_MAX];
> +
> +int cfs_tracefile_init_arch(void)
> +{
> + int i;
> + int j;
> + struct cfs_trace_cpu_data *tcd;
> +
> + /* initialize trace_data */
> + memset(cfs_trace_data, 0, sizeof(cfs_trace_data));
> + for (i = 0; i < CFS_TCD_TYPE_MAX; i++) {
> + cfs_trace_data[i] =
> + kmalloc_array(num_possible_cpus(),
> + sizeof(union cfs_trace_data_union),
> + GFP_KERNEL);
> + if (!cfs_trace_data[i])
> + goto out;
> + }
> +
> + /* arch related info initialized */
> + cfs_tcd_for_each(tcd, i, j) {
> + spin_lock_init(&tcd->tcd_lock);
> + tcd->tcd_pages_factor = pages_factor[i];
> + tcd->tcd_type = i;
> + tcd->tcd_cpu = j;
> + }
> +
> + for (i = 0; i < num_possible_cpus(); i++)
> + for (j = 0; j < 3; j++) {
> + cfs_trace_console_buffers[i][j] =
> + kmalloc(CFS_TRACE_CONSOLE_BUFFER_SIZE,
> + GFP_KERNEL);
> +
> + if (!cfs_trace_console_buffers[i][j])
> + goto out;
> + }
> +
> + return 0;
> +
> +out:
> + cfs_tracefile_fini_arch();
> + pr_err("lnet: Not enough memory\n");
> + return -ENOMEM;
> +}
> +
> +void cfs_tracefile_fini_arch(void)
> +{
> + int i;
> + int j;
> +
> + for (i = 0; i < num_possible_cpus(); i++)
> + for (j = 0; j < 3; j++) {
> + kfree(cfs_trace_console_buffers[i][j]);
> + cfs_trace_console_buffers[i][j] = NULL;
> + }
> +
> + for (i = 0; cfs_trace_data[i]; i++) {
> + kfree(cfs_trace_data[i]);
> + cfs_trace_data[i] = NULL;
> + }
> +}
> +
> +enum cfs_trace_buf_type cfs_trace_buf_idx_get(void)
> +{
> + if (in_irq())
> + return CFS_TCD_TYPE_IRQ;
> + if (in_softirq())
> + return CFS_TCD_TYPE_SOFTIRQ;
> + return CFS_TCD_TYPE_PROC;
> +}
> +
> +void
> +cfs_set_ptldebug_header(struct ptldebug_header *header,
> + struct libcfs_debug_msg_data *msgdata,
> + unsigned long stack)
> +{
> + struct timespec64 ts;
> +
> + ktime_get_real_ts64(&ts);
> +
> + header->ph_subsys = msgdata->msg_subsys;
> + header->ph_mask = msgdata->msg_mask;
> + header->ph_cpu_id = smp_processor_id();
> + header->ph_type = cfs_trace_buf_idx_get();
> + /* y2038 safe since all user space treats this as unsigned, but
> + * will overflow in 2106
> + */
> + header->ph_sec = (u32)ts.tv_sec;
> + header->ph_usec = ts.tv_nsec / NSEC_PER_USEC;
> + header->ph_stack = stack;
> + header->ph_pid = current->pid;
> + header->ph_line_num = msgdata->msg_line;
> + header->ph_extern_pid = 0;
> +}
> +
> +static char *
> +dbghdr_to_err_string(struct ptldebug_header *hdr)
> +{
> + switch (hdr->ph_subsys) {
> + case S_LND:
> + case S_LNET:
> + return "LNetError";
> + default:
> + return "LustreError";
> + }
> +}
> +
> +static char *
> +dbghdr_to_info_string(struct ptldebug_header *hdr)
> +{
> + switch (hdr->ph_subsys) {
> + case S_LND:
> + case S_LNET:
> + return "LNet";
> + default:
> + return "Lustre";
> + }
> +}
> +
> +void cfs_print_to_console(struct ptldebug_header *hdr, int mask,
> + const char *buf, int len, const char *file,
> + const char *fn)
> +{
> + char *prefix = "Lustre", *ptype = NULL;
> +
> + if (mask & D_EMERG) {
> + prefix = dbghdr_to_err_string(hdr);
> + ptype = KERN_EMERG;
> + } else if (mask & D_ERROR) {
> + prefix = dbghdr_to_err_string(hdr);
> + ptype = KERN_ERR;
> + } else if (mask & D_WARNING) {
> + prefix = dbghdr_to_info_string(hdr);
> + ptype = KERN_WARNING;
> + } else if (mask & (D_CONSOLE | libcfs_printk)) {
> + prefix = dbghdr_to_info_string(hdr);
> + ptype = KERN_INFO;
> + }
> +
> + if (mask & D_CONSOLE) {
> + pr_info("%s%s: %.*s", ptype, prefix, len, buf);
> + } else {
> + pr_info("%s%s: %d:%d:(%s:%d:%s()) %.*s", ptype, prefix,
> + hdr->ph_pid, hdr->ph_extern_pid, file,
> + hdr->ph_line_num, fn, len, buf);
> + }
> +}
> +
> static inline struct cfs_trace_page *
> cfs_tage_from_list(struct list_head *list)
> {
>
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES
2018-06-15 7:11 ` [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES NeilBrown
@ 2018-06-21 1:38 ` James Simmons
2018-06-21 4:28 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-21 1:38 UTC (permalink / raw)
To: lustre-devel
> As well as CFS_TCD_TYPE_CNT we have TCD_MAX_TYPES which has a larger
> value but a similar meaning. Discard it and just use
> CFS_TCD_TYPE_CNT.
>
> Two places relied on the fact that TCD_MAX_TYPES was larger and so
> there would be NULLs at the end of the array. Change
> them to check the array size properly.
>
> Signed-off-by: NeilBrown <neilb@suse.com>
Nak: This one really makes my nodes go up in flames when I mount a
file system.
> ---
> drivers/staging/lustre/lnet/libcfs/tracefile.c | 8 ++++----
> drivers/staging/lustre/lnet/libcfs/tracefile.h | 3 +--
> 2 files changed, 5 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
> index cdef67391a72..555ade0ddafb 100644
> --- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
> +++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
> @@ -50,7 +50,7 @@
> #include "tracefile.h"
>
> /* XXX move things up to the top, comment */
> -union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS] __cacheline_aligned;
> +union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS] __cacheline_aligned;
>
> char cfs_tracefile[TRACEFILE_NAME_SIZE];
> long long cfs_tracefile_size = CFS_TRACEFILE_SIZE;
> @@ -145,8 +145,8 @@ void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
> spin_unlock(&tcd->tcd_lock);
> }
>
> -#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
> - for (i = 0; cfs_trace_data[i] && \
> +#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i] && \
> (tcd = &(*cfs_trace_data[i])[cpu].tcd) && \
> cfs_trace_lock_tcd(tcd, 1); cfs_trace_unlock_tcd(tcd, 1), i++)
>
> @@ -1381,7 +1381,7 @@ static void cfs_trace_cleanup(void)
> cfs_trace_console_buffers[i][j] = NULL;
> }
>
> - for (i = 0; cfs_trace_data[i]; i++) {
> + for (i = 0; cfs_trace_data[i] && i < CFS_TCD_TYPE_CNT; i++) {
> kfree(cfs_trace_data[i]);
> cfs_trace_data[i] = NULL;
> }
> diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
> index 23faecf886c1..a09d5b9349f0 100644
> --- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
> +++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
> @@ -184,8 +184,7 @@ union cfs_trace_data_union {
> char __pad[L1_CACHE_ALIGN(sizeof(struct cfs_trace_cpu_data))];
> };
>
> -#define TCD_MAX_TYPES 8
> -extern union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS];
> +extern union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS];
>
> #define cfs_tcd_for_each(tcd, i, j) \
> for (i = 0; cfs_trace_data[i]; i++) \
>
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
` (23 preceding siblings ...)
2018-06-15 7:11 ` [lustre-devel] [PATCH 23/24] lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT NeilBrown
@ 2018-06-21 1:48 ` James Simmons
2018-06-21 2:29 ` Patrick Farrell
2018-06-21 2:54 ` NeilBrown
24 siblings, 2 replies; 82+ messages in thread
From: James Simmons @ 2018-06-21 1:48 UTC (permalink / raw)
To: lustre-devel
> Here are a bunch more cleanups that will appear in my lustre-testing
> tree shortly.
>
> The interesting stuff is some enhancements to kbuild to let us
> build modules from multiple directories.
> I've used that to clean up the build of ptlrpc+ldlm and to
> merge libcfs and lnet into the one module - I've left the lnds
> separate.
That looks pretty good. It doesn't apply to the stable lustre tree but I
did try it out on lustre-testing. The module building does seem to
function well. Only thing not clear is if do a
make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
do lmv.a and the other *.a files get relinked int lustre.ko ?
> I've haven't progressed further in lustre because I'm not sure which -
> if any - modules are also used by the server.
> I'd rather shared code was in a separate module to client-only.
>
> Which parts of lustre/lustre/* are also used by server?
Basically all the client code is reused on the server except for llite.
The reason is each server type MDS, MGS, or OSS is treated as a client
to each other. So in affect the server code is a middle layer level on
top of the client code.
> After that, it is all fairly simple cleanups.
>
> I plan to submit the kbuild stuff upstream next week.
>
> Thanks,
> NeilBrown
>
> ---
>
> NeilBrown (24):
> kbuild: detect directories in components of a module.
> kbuild: treat a directory listed in a composite object as foo/mod.a
> kbuild: support building of per-directory mod.a
> kbuild: disable KBUILD_MODNAME when building for mod.a
> kbuild: Add documentation for modobj-m
> lustre: build ldlm in the ldlm directory.
> lustre: merge libcfs module into lnet.
This seems to work when I tried the lustre-testing tree. I did see
regression to related to this work in that tree.
> lustre: use truncate_inode_page in place of truncate_complete_page
> lustre: don't use spare bits in iattr.ia_valid
> lustre: remove lustre_patchless_compat.h
> lustre: discard current_n*groups macros.
> lustre: discard LTIME_S macro
> lustre: discard NO_QUOTA and QUOTA_OK
> lustre: discard ext2* bit operations
> lustre: centralize TIMES_SET_FLAGS
> lustre: remove redefinition of module_init().
> lustre: remove lustre_compat.h
The first two patches of the above seem to break things. The first one was
pointed out so I avoid issues in my testing by changing the patch to just
move truncate_inode_page() to lustre_mdc.h for now. The next patch does
break things.
> lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
> lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
> lustre: move tcd locking across to tracefile.c
> lustre: move remainder of linux-tracefile.c to tracefile.c
> lustre: fold cfs_tracefile_*_arch into their only callers.
> lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
> lustre: discard TCD_MAX_TYPES
I see breakage in this code as well.
> Documentation/kbuild/makefiles.txt | 65 +++++
> drivers/staging/lustre/lnet/Makefile | 7 -
> drivers/staging/lustre/lnet/libcfs/Makefile | 7 -
> drivers/staging/lustre/lnet/libcfs/debug.c | 6
> .../staging/lustre/lnet/libcfs/linux-tracefile.c | 258 --------------------
> drivers/staging/lustre/lnet/libcfs/module.c | 26 +-
> drivers/staging/lustre/lnet/libcfs/tracefile.c | 241 +++++++++++++++++--
> drivers/staging/lustre/lnet/libcfs/tracefile.h | 40 ---
> drivers/staging/lustre/lnet/lnet/Makefile | 11 -
> drivers/staging/lustre/lnet/lnet/module.c | 9 +
> drivers/staging/lustre/lustre/fid/lproc_fid.c | 1
> drivers/staging/lustre/lustre/fld/lproc_fld.c | 1
> drivers/staging/lustre/lustre/include/cl_object.h | 6
> .../staging/lustre/lustre/include/lustre_compat.h | 82 ------
> drivers/staging/lustre/lustre/include/lustre_net.h | 1
> .../lustre/include/lustre_patchless_compat.h | 68 -----
> drivers/staging/lustre/lustre/include/obd.h | 7 +
> .../staging/lustre/lustre/include/obd_support.h | 1
> drivers/staging/lustre/lustre/ldlm/Makefile | 11 +
> drivers/staging/lustre/lustre/llite/dir.c | 3
> drivers/staging/lustre/lustre/llite/file.c | 36 +--
> drivers/staging/lustre/lustre/llite/lcommon_cl.c | 11 -
> .../staging/lustre/lustre/llite/llite_internal.h | 13 +
> drivers/staging/lustre/lustre/llite/llite_lib.c | 52 ++--
> drivers/staging/lustre/lustre/llite/namei.c | 16 +
> drivers/staging/lustre/lustre/llite/vvp_io.c | 8 -
> drivers/staging/lustre/lustre/llite/vvp_object.c | 6
> drivers/staging/lustre/lustre/lmv/lmv_intent.c | 6
> drivers/staging/lustre/lustre/lmv/lmv_obd.c | 21 +-
> drivers/staging/lustre/lustre/lov/lov_io.c | 3
> drivers/staging/lustre/lustre/lov/lov_obd.c | 2
> drivers/staging/lustre/lustre/lov/lov_pack.c | 1
> drivers/staging/lustre/lustre/lov/lov_page.c | 1
> drivers/staging/lustre/lustre/mdc/mdc_lib.c | 19 +
> drivers/staging/lustre/lustre/mdc/mdc_reint.c | 4
> drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 -
> drivers/staging/lustre/lustre/obdclass/class_obd.c | 1
> drivers/staging/lustre/lustre/obdclass/llog.c | 7 -
> .../lustre/lustre/obdclass/lprocfs_status.c | 1
> .../staging/lustre/lustre/obdclass/obd_config.c | 1
> drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1
> drivers/staging/lustre/lustre/obdclass/obdo.c | 10 -
> .../staging/lustre/lustre/obdecho/echo_client.c | 1
> drivers/staging/lustre/lustre/osc/lproc_osc.c | 1
> drivers/staging/lustre/lustre/osc/osc_cache.c | 4
> drivers/staging/lustre/lustre/osc/osc_io.c | 23 +-
> drivers/staging/lustre/lustre/osc/osc_quota.c | 4
> drivers/staging/lustre/lustre/osc/osc_request.c | 1
> drivers/staging/lustre/lustre/ptlrpc/Makefile | 12 -
> drivers/staging/lustre/lustre/ptlrpc/import.c | 1
> .../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c | 1
> drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c | 1
> drivers/staging/lustre/lustre/ptlrpc/sec.c | 15 +
> drivers/staging/lustre/lustre/ptlrpc/sec_plain.c | 1
> drivers/staging/lustre/lustre/ptlrpc/service.c | 1
> scripts/Makefile.build | 57 +++-
> scripts/Makefile.lib | 63 +++--
> 57 files changed, 584 insertions(+), 685 deletions(-)
> delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
>
> --
> Signature
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 1:48 ` [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction James Simmons
@ 2018-06-21 2:29 ` Patrick Farrell
2018-06-21 2:57 ` James Simmons
2018-06-21 3:00 ` NeilBrown
2018-06-21 2:54 ` NeilBrown
1 sibling, 2 replies; 82+ messages in thread
From: Patrick Farrell @ 2018-06-21 2:29 UTC (permalink / raw)
To: lustre-devel
Huh? Very little of the client is used on the server. The vast majority of the client is llite, vvp, then lov, osc, and mdc. These are not used on the servers at all, the modules won't even be loaded.
Servers are definitely *not* clients of one another in the sense of "Lustre clients". They interact with each other, but in a manner very different from client to server.
The ldlm and ptlrpc layers are *partly* shared, and of course, the networking from lnet down is. But that's not "the client".
The statement that the server layer is a middle layer on top of the client code doesn't make any sense to me. It's like this:
vfs
client code
ptlrpc
networking
------ PHYSICAL BOUNDARY --------
networking
ptlrpc
server code
ldiskfs
with LDLM also being explicitly shared, though client & server code is not all shared. Very little to nothing outside of the networking & ldlm layers is shared.
________________________________
From: lustre-devel <lustre-devel-bounces@lists.lustre.org> on behalf of James Simmons <jsimmons@infradead.org>
Sent: Wednesday, June 20, 2018 8:48:07 PM
To: NeilBrown
Cc: Oleg Drokin; Andreas Dilger; Lustre Development List
Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
> Here are a bunch more cleanups that will appear in my lustre-testing
> tree shortly.
>
> The interesting stuff is some enhancements to kbuild to let us
> build modules from multiple directories.
> I've used that to clean up the build of ptlrpc+ldlm and to
> merge libcfs and lnet into the one module - I've left the lnds
> separate.
That looks pretty good. It doesn't apply to the stable lustre tree but I
did try it out on lustre-testing. The module building does seem to
function well. Only thing not clear is if do a
make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
do lmv.a and the other *.a files get relinked int lustre.ko ?
> I've haven't progressed further in lustre because I'm not sure which -
> if any - modules are also used by the server.
> I'd rather shared code was in a separate module to client-only.
>
> Which parts of lustre/lustre/* are also used by server?
Basically all the client code is reused on the server except for llite.
The reason is each server type MDS, MGS, or OSS is treated as a client
to each other. So in affect the server code is a middle layer level on
top of the client code.
> After that, it is all fairly simple cleanups.
>
> I plan to submit the kbuild stuff upstream next week.
>
> Thanks,
> NeilBrown
>
> ---
>
> NeilBrown (24):
> kbuild: detect directories in components of a module.
> kbuild: treat a directory listed in a composite object as foo/mod.a
> kbuild: support building of per-directory mod.a
> kbuild: disable KBUILD_MODNAME when building for mod.a
> kbuild: Add documentation for modobj-m
> lustre: build ldlm in the ldlm directory.
> lustre: merge libcfs module into lnet.
This seems to work when I tried the lustre-testing tree. I did see
regression to related to this work in that tree.
> lustre: use truncate_inode_page in place of truncate_complete_page
> lustre: don't use spare bits in iattr.ia_valid
> lustre: remove lustre_patchless_compat.h
> lustre: discard current_n*groups macros.
> lustre: discard LTIME_S macro
> lustre: discard NO_QUOTA and QUOTA_OK
> lustre: discard ext2* bit operations
> lustre: centralize TIMES_SET_FLAGS
> lustre: remove redefinition of module_init().
> lustre: remove lustre_compat.h
The first two patches of the above seem to break things. The first one was
pointed out so I avoid issues in my testing by changing the patch to just
move truncate_inode_page() to lustre_mdc.h for now. The next patch does
break things.
> lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
> lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
> lustre: move tcd locking across to tracefile.c
> lustre: move remainder of linux-tracefile.c to tracefile.c
> lustre: fold cfs_tracefile_*_arch into their only callers.
> lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
> lustre: discard TCD_MAX_TYPES
I see breakage in this code as well.
> Documentation/kbuild/makefiles.txt | 65 +++++
> drivers/staging/lustre/lnet/Makefile | 7 -
> drivers/staging/lustre/lnet/libcfs/Makefile | 7 -
> drivers/staging/lustre/lnet/libcfs/debug.c | 6
> .../staging/lustre/lnet/libcfs/linux-tracefile.c | 258 --------------------
> drivers/staging/lustre/lnet/libcfs/module.c | 26 +-
> drivers/staging/lustre/lnet/libcfs/tracefile.c | 241 +++++++++++++++++--
> drivers/staging/lustre/lnet/libcfs/tracefile.h | 40 ---
> drivers/staging/lustre/lnet/lnet/Makefile | 11 -
> drivers/staging/lustre/lnet/lnet/module.c | 9 +
> drivers/staging/lustre/lustre/fid/lproc_fid.c | 1
> drivers/staging/lustre/lustre/fld/lproc_fld.c | 1
> drivers/staging/lustre/lustre/include/cl_object.h | 6
> .../staging/lustre/lustre/include/lustre_compat.h | 82 ------
> drivers/staging/lustre/lustre/include/lustre_net.h | 1
> .../lustre/include/lustre_patchless_compat.h | 68 -----
> drivers/staging/lustre/lustre/include/obd.h | 7 +
> .../staging/lustre/lustre/include/obd_support.h | 1
> drivers/staging/lustre/lustre/ldlm/Makefile | 11 +
> drivers/staging/lustre/lustre/llite/dir.c | 3
> drivers/staging/lustre/lustre/llite/file.c | 36 +--
> drivers/staging/lustre/lustre/llite/lcommon_cl.c | 11 -
> .../staging/lustre/lustre/llite/llite_internal.h | 13 +
> drivers/staging/lustre/lustre/llite/llite_lib.c | 52 ++--
> drivers/staging/lustre/lustre/llite/namei.c | 16 +
> drivers/staging/lustre/lustre/llite/vvp_io.c | 8 -
> drivers/staging/lustre/lustre/llite/vvp_object.c | 6
> drivers/staging/lustre/lustre/lmv/lmv_intent.c | 6
> drivers/staging/lustre/lustre/lmv/lmv_obd.c | 21 +-
> drivers/staging/lustre/lustre/lov/lov_io.c | 3
> drivers/staging/lustre/lustre/lov/lov_obd.c | 2
> drivers/staging/lustre/lustre/lov/lov_pack.c | 1
> drivers/staging/lustre/lustre/lov/lov_page.c | 1
> drivers/staging/lustre/lustre/mdc/mdc_lib.c | 19 +
> drivers/staging/lustre/lustre/mdc/mdc_reint.c | 4
> drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 -
> drivers/staging/lustre/lustre/obdclass/class_obd.c | 1
> drivers/staging/lustre/lustre/obdclass/llog.c | 7 -
> .../lustre/lustre/obdclass/lprocfs_status.c | 1
> .../staging/lustre/lustre/obdclass/obd_config.c | 1
> drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1
> drivers/staging/lustre/lustre/obdclass/obdo.c | 10 -
> .../staging/lustre/lustre/obdecho/echo_client.c | 1
> drivers/staging/lustre/lustre/osc/lproc_osc.c | 1
> drivers/staging/lustre/lustre/osc/osc_cache.c | 4
> drivers/staging/lustre/lustre/osc/osc_io.c | 23 +-
> drivers/staging/lustre/lustre/osc/osc_quota.c | 4
> drivers/staging/lustre/lustre/osc/osc_request.c | 1
> drivers/staging/lustre/lustre/ptlrpc/Makefile | 12 -
> drivers/staging/lustre/lustre/ptlrpc/import.c | 1
> .../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c | 1
> drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c | 1
> drivers/staging/lustre/lustre/ptlrpc/sec.c | 15 +
> drivers/staging/lustre/lustre/ptlrpc/sec_plain.c | 1
> drivers/staging/lustre/lustre/ptlrpc/service.c | 1
> scripts/Makefile.build | 57 +++-
> scripts/Makefile.lib | 63 +++--
> 57 files changed, 584 insertions(+), 685 deletions(-)
> delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
>
> --
> Signature
>
>
_______________________________________________
lustre-devel mailing list
lustre-devel at lists.lustre.org
http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/c4685e67/attachment-0001.html>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 1:48 ` [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction James Simmons
2018-06-21 2:29 ` Patrick Farrell
@ 2018-06-21 2:54 ` NeilBrown
2018-06-24 20:46 ` James Simmons
1 sibling, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-21 2:54 UTC (permalink / raw)
To: lustre-devel
On Thu, Jun 21 2018, James Simmons wrote:
>> Here are a bunch more cleanups that will appear in my lustre-testing
>> tree shortly.
>>
>> The interesting stuff is some enhancements to kbuild to let us
>> build modules from multiple directories.
>> I've used that to clean up the build of ptlrpc+ldlm and to
>> merge libcfs and lnet into the one module - I've left the lnds
>> separate.
>
> That looks pretty good. It doesn't apply to the stable lustre tree but I
> did try it out on lustre-testing. The module building does seem to
> function well. Only thing not clear is if do a
>
> make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
Why would you do that? SUBDIRS= is for building external modules.
What are you trying to achieve?
>
> do lmv.a and the other *.a files get relinked int lustre.ko ?
It isn't "lmv.a", it is ".../lustre/lmv/mod.a".
>
>> I've haven't progressed further in lustre because I'm not sure which -
>> if any - modules are also used by the server.
>> I'd rather shared code was in a separate module to client-only.
>>
>> Which parts of lustre/lustre/* are also used by server?
>
> Basically all the client code is reused on the server except for llite.
> The reason is each server type MDS, MGS, or OSS is treated as a client
> to each other. So in affect the server code is a middle layer level on
> top of the client code.
>
>> After that, it is all fairly simple cleanups.
>>
>> I plan to submit the kbuild stuff upstream next week.
>>
>> Thanks,
>> NeilBrown
>>
>> ---
>>
>> NeilBrown (24):
>> kbuild: detect directories in components of a module.
>> kbuild: treat a directory listed in a composite object as foo/mod.a
>> kbuild: support building of per-directory mod.a
>> kbuild: disable KBUILD_MODNAME when building for mod.a
>> kbuild: Add documentation for modobj-m
>> lustre: build ldlm in the ldlm directory.
>> lustre: merge libcfs module into lnet.
>
> This seems to work when I tried the lustre-testing tree. I did see
> regression to related to this work in that tree.
>
>> lustre: use truncate_inode_page in place of truncate_complete_page
>> lustre: don't use spare bits in iattr.ia_valid
>> lustre: remove lustre_patchless_compat.h
>> lustre: discard current_n*groups macros.
>> lustre: discard LTIME_S macro
>> lustre: discard NO_QUOTA and QUOTA_OK
>> lustre: discard ext2* bit operations
>> lustre: centralize TIMES_SET_FLAGS
>> lustre: remove redefinition of module_init().
>> lustre: remove lustre_compat.h
>
> The first two patches of the above seem to break things. The first one was
> pointed out so I avoid issues in my testing by changing the patch to just
> move truncate_inode_page() to lustre_mdc.h for now. The next patch does
> break things.
Can you be specific about what things (or just one thing) that the next
patch breaks? I haven't seen anything in my testing.
The only problem I've seen is that kvmalloc(GFP_NOFS) has been changed to
silently become kmalloc(GFP_NOFS) so there are memory allocation
problems.
>
>> lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
>> lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
>> lustre: move tcd locking across to tracefile.c
>> lustre: move remainder of linux-tracefile.c to tracefile.c
>> lustre: fold cfs_tracefile_*_arch into their only callers.
>> lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
>> lustre: discard TCD_MAX_TYPES
>
> I see breakage in this code as well.
Specifically??...
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/ed188aeb/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 2:29 ` Patrick Farrell
@ 2018-06-21 2:57 ` James Simmons
2018-06-21 7:22 ` Andreas Dilger
2018-06-21 3:00 ` NeilBrown
1 sibling, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-21 2:57 UTC (permalink / raw)
To: lustre-devel
> Huh?? Very little of the client is used on the server.? The vast majority of the client is llite, vvp, then lov,?osc, and?mdc.? These are
> not used on the servers at all, the modules won't even be loaded.
>
>
> Servers are definitely *not* clients of one another in the sense of "Lustre clients".? They interact with each other, but in a manner very
> different from client to server.
>
> The ldlm and ptlrpc layers are *partly* shared, and of course, the networking from lnet down is.? But that's not "the client".
>
> The statement that the server layer is a middle layer on top of the client code doesn't make any sense to me.? It's like this:
>
>
> vfs
> client code
>
> ptlrpc
>
> networking
>
> ------ PHYSICAL BOUNDARY --------
> networking
>
> ptlrpc
>
> server code
>
> ldiskfs
>
>
>
> with LDLM also being explicitly shared, though client & server code is not all shared.? Very little to nothing outside of the networking &
> ldlm layers is shared.
When I first started this (pre 2.3 days) that is how it was explained to
me. I just took it at face value but when it comes to the separation as I
don't have a clue. Patrick since you have a better grasp of the
architecture can you provide details to Neil. Perhaps to documentation on
this separation if it exist.
> > Here are a bunch more cleanups that will appear in my lustre-testing
> > tree shortly.
> >
> > The interesting stuff is some enhancements to kbuild to let us
> > build modules from multiple directories.
> > I've used that to clean up the build of ptlrpc+ldlm and to
> > merge libcfs and lnet into the one module - I've left the lnds
> > separate.
>
> That looks pretty good. It doesn't apply to the stable lustre tree but I
> did try it out on lustre-testing. The module building does seem to
> function well. Only thing not clear is if do a
>
> make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
>
> do lmv.a and the other *.a files get relinked int lustre.ko ?
> ?
> > I've haven't progressed further in lustre because I'm not sure which -
> > if any - modules are also used by the server.
> > I'd rather shared code was in a separate module to client-only.
> >
> > Which parts of lustre/lustre/* are also used by server?
>
> Basically all the client code is reused on the server except for llite.
> The reason is each server type MDS, MGS, or OSS is treated as a client
> to each other. So in affect the server code is a middle layer level on
> top of the client code.
>
> > After that, it is all fairly simple cleanups.
> >
> > I plan to submit the kbuild stuff upstream next week.
> >
> > Thanks,
> > NeilBrown
> >
> > ---
> >
> > NeilBrown (24):
> >?????? kbuild: detect directories in components of a module.
> >?????? kbuild: treat a directory listed in a composite object as foo/mod.a
> >?????? kbuild: support building of per-directory mod.a
> >?????? kbuild: disable KBUILD_MODNAME when building for mod.a
> >?????? kbuild: Add documentation for modobj-m
> >?????? lustre: build ldlm in the ldlm directory.
> >?????? lustre: merge libcfs module into lnet.
>
> This seems to work when I tried the lustre-testing tree. I did see
> regression to related to this work in that tree.
>
> >?????? lustre: use truncate_inode_page in place of truncate_complete_page
> >?????? lustre: don't use spare bits in iattr.ia_valid
> >?????? lustre: remove lustre_patchless_compat.h
> >?????? lustre: discard current_n*groups macros.
> >?????? lustre: discard LTIME_S macro
> >?????? lustre: discard NO_QUOTA and QUOTA_OK
> >?????? lustre: discard ext2* bit operations
> >?????? lustre: centralize TIMES_SET_FLAGS
> >?????? lustre: remove redefinition of module_init().
> >?????? lustre: remove lustre_compat.h
>
> The first two patches of the above seem to break things. The first one was
> pointed out so I avoid issues in my testing by changing the patch to just
> move truncate_inode_page() to lustre_mdc.h for now. The next patch does
> break things.
>
> >?????? lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
> >?????? lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
> >?????? lustre: move tcd locking across to tracefile.c
> >?????? lustre: move remainder of linux-tracefile.c to tracefile.c
> >?????? lustre: fold cfs_tracefile_*_arch into their only callers.
> >?????? lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
> >?????? lustre: discard TCD_MAX_TYPES
>
> I see breakage in this code as well.
>
> >? Documentation/kbuild/makefiles.txt???????????????? |?? 65 +++++
> >? drivers/staging/lustre/lnet/Makefile?????????????? |??? 7 -
> >? drivers/staging/lustre/lnet/libcfs/Makefile??????? |??? 7 -
> >? drivers/staging/lustre/lnet/libcfs/debug.c???????? |??? 6
> >? .../staging/lustre/lnet/libcfs/linux-tracefile.c?? |? 258 --------------------
> >? drivers/staging/lustre/lnet/libcfs/module.c??????? |?? 26 +-
> >? drivers/staging/lustre/lnet/libcfs/tracefile.c???? |? 241 +++++++++++++++++--
> >? drivers/staging/lustre/lnet/libcfs/tracefile.h???? |?? 40 ---
> >? drivers/staging/lustre/lnet/lnet/Makefile????????? |?? 11 -
> >? drivers/staging/lustre/lnet/lnet/module.c????????? |??? 9 +
> >? drivers/staging/lustre/lustre/fid/lproc_fid.c????? |??? 1
> >? drivers/staging/lustre/lustre/fld/lproc_fld.c????? |??? 1
> >? drivers/staging/lustre/lustre/include/cl_object.h? |??? 6
> >? .../staging/lustre/lustre/include/lustre_compat.h? |?? 82 ------
> >? drivers/staging/lustre/lustre/include/lustre_net.h |??? 1
> >? .../lustre/include/lustre_patchless_compat.h?????? |?? 68 -----
> >? drivers/staging/lustre/lustre/include/obd.h??????? |??? 7 +
> >? .../staging/lustre/lustre/include/obd_support.h??? |??? 1
> >? drivers/staging/lustre/lustre/ldlm/Makefile??????? |?? 11 +
> >? drivers/staging/lustre/lustre/llite/dir.c????????? |??? 3
> >? drivers/staging/lustre/lustre/llite/file.c???????? |?? 36 +--
> >? drivers/staging/lustre/lustre/llite/lcommon_cl.c?? |?? 11 -
> >? .../staging/lustre/lustre/llite/llite_internal.h?? |?? 13 +
> >? drivers/staging/lustre/lustre/llite/llite_lib.c??? |?? 52 ++--
> >? drivers/staging/lustre/lustre/llite/namei.c??????? |?? 16 +
> >? drivers/staging/lustre/lustre/llite/vvp_io.c?????? |??? 8 -
> >? drivers/staging/lustre/lustre/llite/vvp_object.c?? |??? 6
> >? drivers/staging/lustre/lustre/lmv/lmv_intent.c???? |??? 6
> >? drivers/staging/lustre/lustre/lmv/lmv_obd.c??????? |?? 21 +-
> >? drivers/staging/lustre/lustre/lov/lov_io.c???????? |??? 3
> >? drivers/staging/lustre/lustre/lov/lov_obd.c??????? |??? 2
> >? drivers/staging/lustre/lustre/lov/lov_pack.c?????? |??? 1
> >? drivers/staging/lustre/lustre/lov/lov_page.c?????? |??? 1
> >? drivers/staging/lustre/lustre/mdc/mdc_lib.c??????? |?? 19 +
> >? drivers/staging/lustre/lustre/mdc/mdc_reint.c????? |??? 4
> >? drivers/staging/lustre/lustre/mdc/mdc_request.c??? |?? 12 -
> >? drivers/staging/lustre/lustre/obdclass/class_obd.c |??? 1
> >? drivers/staging/lustre/lustre/obdclass/llog.c????? |??? 7 -
> >? .../lustre/lustre/obdclass/lprocfs_status.c??????? |??? 1
> >? .../staging/lustre/lustre/obdclass/obd_config.c??? |??? 1
> >? drivers/staging/lustre/lustre/obdclass/obd_mount.c |??? 1
> >? drivers/staging/lustre/lustre/obdclass/obdo.c????? |?? 10 -
> >? .../staging/lustre/lustre/obdecho/echo_client.c??? |??? 1
> >? drivers/staging/lustre/lustre/osc/lproc_osc.c????? |??? 1
> >? drivers/staging/lustre/lustre/osc/osc_cache.c????? |??? 4
> >? drivers/staging/lustre/lustre/osc/osc_io.c???????? |?? 23 +-
> >? drivers/staging/lustre/lustre/osc/osc_quota.c????? |??? 4
> >? drivers/staging/lustre/lustre/osc/osc_request.c??? |??? 1
> >? drivers/staging/lustre/lustre/ptlrpc/Makefile????? |?? 12 -
> >? drivers/staging/lustre/lustre/ptlrpc/import.c????? |??? 1
> >? .../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c??? |??? 1
> >? drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c???? |??? 1
> >? drivers/staging/lustre/lustre/ptlrpc/sec.c???????? |?? 15 +
> >? drivers/staging/lustre/lustre/ptlrpc/sec_plain.c?? |??? 1
> >? drivers/staging/lustre/lustre/ptlrpc/service.c???? |??? 1
> >? scripts/Makefile.build???????????????????????????? |?? 57 +++-
> >? scripts/Makefile.lib?????????????????????????????? |?? 63 +++--
> >? 57 files changed, 584 insertions(+), 685 deletions(-)
> >? delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
> >? delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
> >? delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> >? create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
> >
> > --
> > Signature
> >
> >
> _______________________________________________
> lustre-devel mailing list
> lustre-devel at lists.lustre.org
> http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 2:29 ` Patrick Farrell
2018-06-21 2:57 ` James Simmons
@ 2018-06-21 3:00 ` NeilBrown
2018-06-21 3:42 ` Patrick Farrell
2018-06-24 20:39 ` James Simmons
1 sibling, 2 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-21 3:00 UTC (permalink / raw)
To: lustre-devel
On Thu, Jun 21 2018, Patrick Farrell wrote:
> Huh? Very little of the client is used on the server. The vast majority of the client is llite, vvp, then lov, osc, and mdc. These are not used on the servers at all, the modules won't even be loaded.
>
>
> Servers are definitely *not* clients of one another in the sense of "Lustre clients". They interact with each other, but in a manner very different from client to server.
>
> The ldlm and ptlrpc layers are *partly* shared, and of course, the networking from lnet down is. But that's not "the client".
>
> The statement that the server layer is a middle layer on top of the client code doesn't make any sense to me. It's like this:
So llite fid fld lmv lov mdc mgc osc and obdclass
can all be in a "lustre" module.
ldlm and ptlrpc should be in a separate "ptlrpc" module.
And lnet and the klnds are also separate.
On the server side we would have
lod mdd mdt mgs ofd osp ost
does that seem approximately right?
Thanks,
NeilBrown
>
>
> vfs
> client code
>
> ptlrpc
>
> networking
>
> ------ PHYSICAL BOUNDARY --------
> networking
>
> ptlrpc
>
> server code
>
> ldiskfs
>
>
>
> with LDLM also being explicitly shared, though client & server code is not all shared. Very little to nothing outside of the networking & ldlm layers is shared.
>
> ________________________________
> From: lustre-devel <lustre-devel-bounces@lists.lustre.org> on behalf of James Simmons <jsimmons@infradead.org>
> Sent: Wednesday, June 20, 2018 8:48:07 PM
> To: NeilBrown
> Cc: Oleg Drokin; Andreas Dilger; Lustre Development List
> Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
>
>
>> Here are a bunch more cleanups that will appear in my lustre-testing
>> tree shortly.
>>
>> The interesting stuff is some enhancements to kbuild to let us
>> build modules from multiple directories.
>> I've used that to clean up the build of ptlrpc+ldlm and to
>> merge libcfs and lnet into the one module - I've left the lnds
>> separate.
>
> That looks pretty good. It doesn't apply to the stable lustre tree but I
> did try it out on lustre-testing. The module building does seem to
> function well. Only thing not clear is if do a
>
> make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
>
> do lmv.a and the other *.a files get relinked int lustre.ko ?
>
>> I've haven't progressed further in lustre because I'm not sure which -
>> if any - modules are also used by the server.
>> I'd rather shared code was in a separate module to client-only.
>>
>> Which parts of lustre/lustre/* are also used by server?
>
> Basically all the client code is reused on the server except for llite.
> The reason is each server type MDS, MGS, or OSS is treated as a client
> to each other. So in affect the server code is a middle layer level on
> top of the client code.
>
>> After that, it is all fairly simple cleanups.
>>
>> I plan to submit the kbuild stuff upstream next week.
>>
>> Thanks,
>> NeilBrown
>>
>> ---
>>
>> NeilBrown (24):
>> kbuild: detect directories in components of a module.
>> kbuild: treat a directory listed in a composite object as foo/mod.a
>> kbuild: support building of per-directory mod.a
>> kbuild: disable KBUILD_MODNAME when building for mod.a
>> kbuild: Add documentation for modobj-m
>> lustre: build ldlm in the ldlm directory.
>> lustre: merge libcfs module into lnet.
>
> This seems to work when I tried the lustre-testing tree. I did see
> regression to related to this work in that tree.
>
>> lustre: use truncate_inode_page in place of truncate_complete_page
>> lustre: don't use spare bits in iattr.ia_valid
>> lustre: remove lustre_patchless_compat.h
>> lustre: discard current_n*groups macros.
>> lustre: discard LTIME_S macro
>> lustre: discard NO_QUOTA and QUOTA_OK
>> lustre: discard ext2* bit operations
>> lustre: centralize TIMES_SET_FLAGS
>> lustre: remove redefinition of module_init().
>> lustre: remove lustre_compat.h
>
> The first two patches of the above seem to break things. The first one was
> pointed out so I avoid issues in my testing by changing the patch to just
> move truncate_inode_page() to lustre_mdc.h for now. The next patch does
> break things.
>
>> lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
>> lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
>> lustre: move tcd locking across to tracefile.c
>> lustre: move remainder of linux-tracefile.c to tracefile.c
>> lustre: fold cfs_tracefile_*_arch into their only callers.
>> lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
>> lustre: discard TCD_MAX_TYPES
>
> I see breakage in this code as well.
>
>> Documentation/kbuild/makefiles.txt | 65 +++++
>> drivers/staging/lustre/lnet/Makefile | 7 -
>> drivers/staging/lustre/lnet/libcfs/Makefile | 7 -
>> drivers/staging/lustre/lnet/libcfs/debug.c | 6
>> .../staging/lustre/lnet/libcfs/linux-tracefile.c | 258 --------------------
>> drivers/staging/lustre/lnet/libcfs/module.c | 26 +-
>> drivers/staging/lustre/lnet/libcfs/tracefile.c | 241 +++++++++++++++++--
>> drivers/staging/lustre/lnet/libcfs/tracefile.h | 40 ---
>> drivers/staging/lustre/lnet/lnet/Makefile | 11 -
>> drivers/staging/lustre/lnet/lnet/module.c | 9 +
>> drivers/staging/lustre/lustre/fid/lproc_fid.c | 1
>> drivers/staging/lustre/lustre/fld/lproc_fld.c | 1
>> drivers/staging/lustre/lustre/include/cl_object.h | 6
>> .../staging/lustre/lustre/include/lustre_compat.h | 82 ------
>> drivers/staging/lustre/lustre/include/lustre_net.h | 1
>> .../lustre/include/lustre_patchless_compat.h | 68 -----
>> drivers/staging/lustre/lustre/include/obd.h | 7 +
>> .../staging/lustre/lustre/include/obd_support.h | 1
>> drivers/staging/lustre/lustre/ldlm/Makefile | 11 +
>> drivers/staging/lustre/lustre/llite/dir.c | 3
>> drivers/staging/lustre/lustre/llite/file.c | 36 +--
>> drivers/staging/lustre/lustre/llite/lcommon_cl.c | 11 -
>> .../staging/lustre/lustre/llite/llite_internal.h | 13 +
>> drivers/staging/lustre/lustre/llite/llite_lib.c | 52 ++--
>> drivers/staging/lustre/lustre/llite/namei.c | 16 +
>> drivers/staging/lustre/lustre/llite/vvp_io.c | 8 -
>> drivers/staging/lustre/lustre/llite/vvp_object.c | 6
>> drivers/staging/lustre/lustre/lmv/lmv_intent.c | 6
>> drivers/staging/lustre/lustre/lmv/lmv_obd.c | 21 +-
>> drivers/staging/lustre/lustre/lov/lov_io.c | 3
>> drivers/staging/lustre/lustre/lov/lov_obd.c | 2
>> drivers/staging/lustre/lustre/lov/lov_pack.c | 1
>> drivers/staging/lustre/lustre/lov/lov_page.c | 1
>> drivers/staging/lustre/lustre/mdc/mdc_lib.c | 19 +
>> drivers/staging/lustre/lustre/mdc/mdc_reint.c | 4
>> drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 -
>> drivers/staging/lustre/lustre/obdclass/class_obd.c | 1
>> drivers/staging/lustre/lustre/obdclass/llog.c | 7 -
>> .../lustre/lustre/obdclass/lprocfs_status.c | 1
>> .../staging/lustre/lustre/obdclass/obd_config.c | 1
>> drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1
>> drivers/staging/lustre/lustre/obdclass/obdo.c | 10 -
>> .../staging/lustre/lustre/obdecho/echo_client.c | 1
>> drivers/staging/lustre/lustre/osc/lproc_osc.c | 1
>> drivers/staging/lustre/lustre/osc/osc_cache.c | 4
>> drivers/staging/lustre/lustre/osc/osc_io.c | 23 +-
>> drivers/staging/lustre/lustre/osc/osc_quota.c | 4
>> drivers/staging/lustre/lustre/osc/osc_request.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/Makefile | 12 -
>> drivers/staging/lustre/lustre/ptlrpc/import.c | 1
>> .../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/sec.c | 15 +
>> drivers/staging/lustre/lustre/ptlrpc/sec_plain.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/service.c | 1
>> scripts/Makefile.build | 57 +++-
>> scripts/Makefile.lib | 63 +++--
>> 57 files changed, 584 insertions(+), 685 deletions(-)
>> delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
>> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
>> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
>> create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
>>
>> --
>> Signature
>>
>>
> _______________________________________________
> lustre-devel mailing list
> lustre-devel at lists.lustre.org
> http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/c2bc7f0e/attachment-0001.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 3:00 ` NeilBrown
@ 2018-06-21 3:42 ` Patrick Farrell
2018-06-24 20:39 ` James Simmons
1 sibling, 0 replies; 82+ messages in thread
From: Patrick Farrell @ 2018-06-21 3:42 UTC (permalink / raw)
To: lustre-devel
I?m hoping someone with a better memory or who?s in a position to check this a bit better (no computer access this second) will speak up, but...
To a first approximation, yes, that sounds close. Probably missing something but...
Also, not sure about fid and fld. Might be needed on the server...? But you will of course find you when you try.
________________________________
From: NeilBrown <neilb@suse.com>
Sent: Wednesday, June 20, 2018 10:00:54 PM
To: Patrick Farrell; James Simmons
Cc: Lustre Development List
Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
On Thu, Jun 21 2018, Patrick Farrell wrote:
> Huh? Very little of the client is used on the server. The vast majority of the client is llite, vvp, then lov, osc, and mdc. These are not used on the servers at all, the modules won't even be loaded.
>
>
> Servers are definitely *not* clients of one another in the sense of "Lustre clients". They interact with each other, but in a manner very different from client to server.
>
> The ldlm and ptlrpc layers are *partly* shared, and of course, the networking from lnet down is. But that's not "the client".
>
> The statement that the server layer is a middle layer on top of the client code doesn't make any sense to me. It's like this:
So llite fid fld lmv lov mdc mgc osc and obdclass
can all be in a "lustre" module.
ldlm and ptlrpc should be in a separate "ptlrpc" module.
And lnet and the klnds are also separate.
On the server side we would have
lod mdd mdt mgs ofd osp ost
does that seem approximately right?
Thanks,
NeilBrown
>
>
> vfs
> client code
>
> ptlrpc
>
> networking
>
> ------ PHYSICAL BOUNDARY --------
> networking
>
> ptlrpc
>
> server code
>
> ldiskfs
>
>
>
> with LDLM also being explicitly shared, though client & server code is not all shared. Very little to nothing outside of the networking & ldlm layers is shared.
>
> ________________________________
> From: lustre-devel <lustre-devel-bounces@lists.lustre.org> on behalf of James Simmons <jsimmons@infradead.org>
> Sent: Wednesday, June 20, 2018 8:48:07 PM
> To: NeilBrown
> Cc: Oleg Drokin; Andreas Dilger; Lustre Development List
> Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
>
>
>> Here are a bunch more cleanups that will appear in my lustre-testing
>> tree shortly.
>>
>> The interesting stuff is some enhancements to kbuild to let us
>> build modules from multiple directories.
>> I've used that to clean up the build of ptlrpc+ldlm and to
>> merge libcfs and lnet into the one module - I've left the lnds
>> separate.
>
> That looks pretty good. It doesn't apply to the stable lustre tree but I
> did try it out on lustre-testing. The module building does seem to
> function well. Only thing not clear is if do a
>
> make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
>
> do lmv.a and the other *.a files get relinked int lustre.ko ?
>
>> I've haven't progressed further in lustre because I'm not sure which -
>> if any - modules are also used by the server.
>> I'd rather shared code was in a separate module to client-only.
>>
>> Which parts of lustre/lustre/* are also used by server?
>
> Basically all the client code is reused on the server except for llite.
> The reason is each server type MDS, MGS, or OSS is treated as a client
> to each other. So in affect the server code is a middle layer level on
> top of the client code.
>
>> After that, it is all fairly simple cleanups.
>>
>> I plan to submit the kbuild stuff upstream next week.
>>
>> Thanks,
>> NeilBrown
>>
>> ---
>>
>> NeilBrown (24):
>> kbuild: detect directories in components of a module.
>> kbuild: treat a directory listed in a composite object as foo/mod.a
>> kbuild: support building of per-directory mod.a
>> kbuild: disable KBUILD_MODNAME when building for mod.a
>> kbuild: Add documentation for modobj-m
>> lustre: build ldlm in the ldlm directory.
>> lustre: merge libcfs module into lnet.
>
> This seems to work when I tried the lustre-testing tree. I did see
> regression to related to this work in that tree.
>
>> lustre: use truncate_inode_page in place of truncate_complete_page
>> lustre: don't use spare bits in iattr.ia_valid
>> lustre: remove lustre_patchless_compat.h
>> lustre: discard current_n*groups macros.
>> lustre: discard LTIME_S macro
>> lustre: discard NO_QUOTA and QUOTA_OK
>> lustre: discard ext2* bit operations
>> lustre: centralize TIMES_SET_FLAGS
>> lustre: remove redefinition of module_init().
>> lustre: remove lustre_compat.h
>
> The first two patches of the above seem to break things. The first one was
> pointed out so I avoid issues in my testing by changing the patch to just
> move truncate_inode_page() to lustre_mdc.h for now. The next patch does
> break things.
>
>> lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c
>> lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb()
>> lustre: move tcd locking across to tracefile.c
>> lustre: move remainder of linux-tracefile.c to tracefile.c
>> lustre: fold cfs_tracefile_*_arch into their only callers.
>> lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT
>> lustre: discard TCD_MAX_TYPES
>
> I see breakage in this code as well.
>
>> Documentation/kbuild/makefiles.txt | 65 +++++
>> drivers/staging/lustre/lnet/Makefile | 7 -
>> drivers/staging/lustre/lnet/libcfs/Makefile | 7 -
>> drivers/staging/lustre/lnet/libcfs/debug.c | 6
>> .../staging/lustre/lnet/libcfs/linux-tracefile.c | 258 --------------------
>> drivers/staging/lustre/lnet/libcfs/module.c | 26 +-
>> drivers/staging/lustre/lnet/libcfs/tracefile.c | 241 +++++++++++++++++--
>> drivers/staging/lustre/lnet/libcfs/tracefile.h | 40 ---
>> drivers/staging/lustre/lnet/lnet/Makefile | 11 -
>> drivers/staging/lustre/lnet/lnet/module.c | 9 +
>> drivers/staging/lustre/lustre/fid/lproc_fid.c | 1
>> drivers/staging/lustre/lustre/fld/lproc_fld.c | 1
>> drivers/staging/lustre/lustre/include/cl_object.h | 6
>> .../staging/lustre/lustre/include/lustre_compat.h | 82 ------
>> drivers/staging/lustre/lustre/include/lustre_net.h | 1
>> .../lustre/include/lustre_patchless_compat.h | 68 -----
>> drivers/staging/lustre/lustre/include/obd.h | 7 +
>> .../staging/lustre/lustre/include/obd_support.h | 1
>> drivers/staging/lustre/lustre/ldlm/Makefile | 11 +
>> drivers/staging/lustre/lustre/llite/dir.c | 3
>> drivers/staging/lustre/lustre/llite/file.c | 36 +--
>> drivers/staging/lustre/lustre/llite/lcommon_cl.c | 11 -
>> .../staging/lustre/lustre/llite/llite_internal.h | 13 +
>> drivers/staging/lustre/lustre/llite/llite_lib.c | 52 ++--
>> drivers/staging/lustre/lustre/llite/namei.c | 16 +
>> drivers/staging/lustre/lustre/llite/vvp_io.c | 8 -
>> drivers/staging/lustre/lustre/llite/vvp_object.c | 6
>> drivers/staging/lustre/lustre/lmv/lmv_intent.c | 6
>> drivers/staging/lustre/lustre/lmv/lmv_obd.c | 21 +-
>> drivers/staging/lustre/lustre/lov/lov_io.c | 3
>> drivers/staging/lustre/lustre/lov/lov_obd.c | 2
>> drivers/staging/lustre/lustre/lov/lov_pack.c | 1
>> drivers/staging/lustre/lustre/lov/lov_page.c | 1
>> drivers/staging/lustre/lustre/mdc/mdc_lib.c | 19 +
>> drivers/staging/lustre/lustre/mdc/mdc_reint.c | 4
>> drivers/staging/lustre/lustre/mdc/mdc_request.c | 12 -
>> drivers/staging/lustre/lustre/obdclass/class_obd.c | 1
>> drivers/staging/lustre/lustre/obdclass/llog.c | 7 -
>> .../lustre/lustre/obdclass/lprocfs_status.c | 1
>> .../staging/lustre/lustre/obdclass/obd_config.c | 1
>> drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1
>> drivers/staging/lustre/lustre/obdclass/obdo.c | 10 -
>> .../staging/lustre/lustre/obdecho/echo_client.c | 1
>> drivers/staging/lustre/lustre/osc/lproc_osc.c | 1
>> drivers/staging/lustre/lustre/osc/osc_cache.c | 4
>> drivers/staging/lustre/lustre/osc/osc_io.c | 23 +-
>> drivers/staging/lustre/lustre/osc/osc_quota.c | 4
>> drivers/staging/lustre/lustre/osc/osc_request.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/Makefile | 12 -
>> drivers/staging/lustre/lustre/ptlrpc/import.c | 1
>> .../staging/lustre/lustre/ptlrpc/lproc_ptlrpc.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/ptlrpcd.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/sec.c | 15 +
>> drivers/staging/lustre/lustre/ptlrpc/sec_plain.c | 1
>> drivers/staging/lustre/lustre/ptlrpc/service.c | 1
>> scripts/Makefile.build | 57 +++-
>> scripts/Makefile.lib | 63 +++--
>> 57 files changed, 584 insertions(+), 685 deletions(-)
>> delete mode 100644 drivers/staging/lustre/lnet/libcfs/linux-tracefile.c
>> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_compat.h
>> delete mode 100644 drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
>> create mode 100644 drivers/staging/lustre/lustre/ldlm/Makefile
>>
>> --
>> Signature
>>
>>
> _______________________________________________
> lustre-devel mailing list
> lustre-devel at lists.lustre.org
> http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/75c72d16/attachment-0001.html>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid
2018-06-21 1:35 ` James Simmons
@ 2018-06-21 4:19 ` NeilBrown
2018-06-22 2:23 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-21 4:19 UTC (permalink / raw)
To: lustre-devel
On Thu, Jun 21 2018, James Simmons wrote:
>> Using spare bits in iattr.ia_valid is not safe as the
>> bits might get used for some other purpose without
>> us noticing.
>> lustre currently used 4 spare bit, but they are all
>> defined in different places so this isn't immediately
>> obvious.
>>
>> This patch brings all those bit definitions together
>> and defined a new op_data field (op_xvalid) to hold
>> the extra validity bits.
>>
>> It also replaces sa_valid in struct cl_setattr_io
>> with sa_avalid and sa_xvalid. Changing the name is
>> helpful as sa_valid already has another use within
>> lustre.
>>
>> Signed-off-by: NeilBrown <neilb@suse.com>
>
> Nak: I see regressions with this patch.
>
> [12368.453655] Lustre: DEBUG MARKER: == sanity test 240: race between ldlm
> enqueue and the connection RPC (no ASSERT) ===================== 21:
> 16:30 (1529543790)
> [12368.760832] BUG: Dentry
> 000000002646a847{i=200004282000008,n=f237.sanity} still in use (1)
> [unmount of lustre lustre]
> [12368.773746] WARNING: CPU: 1 PID: 10861 at fs/dcache.c:1514
...
> [12369.385564] Lustre: Unmounted lustre-client
> [12369.393247] VFS: Busy inodes after unmount of lustre. Self-destruct in
> 5 seconds. Have a nice day...
>
> When I remove this patch things go back to normal. This will not show up
> if you do a ONLY="240" sh ./sanity.sh. You have to run the sanity.sh in
> total to make this show up.
>
(clearly I read my email in the wrong order)
Very odd. This suggests some sort of life-time-management problem
with inodes, but the changes shouldn't affect that at all.
I've gone back over the patch closely and cannot see anything wrong - it
is really very simple: it just moves flag bits around.
I haven't got as far as 240 yet as changes to kvmalloc have caused
earlier problem (my VMs don't have much RAM). I'll see if I can
coax it all the way to 240 and see what happens.
Thanks for the report.
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/0a7c0f77/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES
2018-06-21 1:38 ` James Simmons
@ 2018-06-21 4:28 ` NeilBrown
2018-06-24 20:37 ` James Simmons
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-21 4:28 UTC (permalink / raw)
To: lustre-devel
As well as CFS_TCD_TYPE_CNT we have TCD_MAX_TYPES which has a larger
value but a similar meaning. Discard it and just use
CFS_TCD_TYPE_CNT.
Two places relied on the fact that TCD_MAX_TYPES was larger and so
there would be NULLs at the end of the array. Change
them to check the array size properly.
Signed-off-by: NeilBrown <neilb@suse.com>
---
Thanks for testing James!
I found two problems.
1/ I had
- for (i = 0; cfs_trace_data[i] && i < CFS_TCD_TYPE_CNT; i++) {
instead of
+ for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) {
So it could dereference beyond the end of an array. I don't this was
the problem.
2/ I hadn't changed
- for (i = 0; cfs_trace_data[i]; i++) \
to
+ for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) \
so if cfs_trace_data[4] was non NULL, bad things could happen.
I suspect this is what happened to you.
Thanks,
NeilBrown
drivers/staging/lustre/lnet/libcfs/tracefile.c | 8 ++++----
drivers/staging/lustre/lnet/libcfs/tracefile.h | 5 ++---
2 files changed, 6 insertions(+), 7 deletions(-)
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
index cdef67391a72..cc399d580444 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
@@ -50,7 +50,7 @@
#include "tracefile.h"
/* XXX move things up to the top, comment */
-union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS] __cacheline_aligned;
+union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS] __cacheline_aligned;
char cfs_tracefile[TRACEFILE_NAME_SIZE];
long long cfs_tracefile_size = CFS_TRACEFILE_SIZE;
@@ -145,8 +145,8 @@ void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
spin_unlock(&tcd->tcd_lock);
}
-#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
- for (i = 0; cfs_trace_data[i] && \
+#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
+ for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i] && \
(tcd = &(*cfs_trace_data[i])[cpu].tcd) && \
cfs_trace_lock_tcd(tcd, 1); cfs_trace_unlock_tcd(tcd, 1), i++)
@@ -1381,7 +1381,7 @@ static void cfs_trace_cleanup(void)
cfs_trace_console_buffers[i][j] = NULL;
}
- for (i = 0; cfs_trace_data[i]; i++) {
+ for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) {
kfree(cfs_trace_data[i]);
cfs_trace_data[i] = NULL;
}
diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
index 23faecf886c1..87b00fc70b70 100644
--- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
+++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
@@ -184,11 +184,10 @@ union cfs_trace_data_union {
char __pad[L1_CACHE_ALIGN(sizeof(struct cfs_trace_cpu_data))];
};
-#define TCD_MAX_TYPES 8
-extern union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS];
+extern union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS];
#define cfs_tcd_for_each(tcd, i, j) \
- for (i = 0; cfs_trace_data[i]; i++) \
+ for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) \
for (j = 0, ((tcd) = &(*cfs_trace_data[i])[j].tcd); \
j < num_possible_cpus(); \
j++, (tcd) = &(*cfs_trace_data[i])[j].tcd)
--
2.14.0.rc0.dirty
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/7a937256/attachment.sig>
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 2:57 ` James Simmons
@ 2018-06-21 7:22 ` Andreas Dilger
2018-06-21 12:28 ` Patrick Farrell
0 siblings, 1 reply; 82+ messages in thread
From: Andreas Dilger @ 2018-06-21 7:22 UTC (permalink / raw)
To: lustre-devel
On Jun 20, 2018, at 8:57 PM, James Simmons <jsimmons@infradead.org> wrote:
>
>>
>> Huh? Very little of the client is used on the server. The vast majority of the client is llite, vvp, then lov, osc, and mdc. These are
>> not used on the servers at all, the modules won't even be loaded.
>>
>>
>> Servers are definitely *not* clients of one another in the sense of "Lustre clients". They interact with each other, but in a manner very
>> different from client to server.
>>
>> The ldlm and ptlrpc layers are *partly* shared, and of course, the networking from lnet down is. But that's not "the client".
>>
>> The statement that the server layer is a middle layer on top of the client code doesn't make any sense to me. It's like this:
>>
>>
>> vfs
>> client code
>> ptlrpc
>> networking
>> ------ PHYSICAL BOUNDARY --------
>> networking
>> ptlrpc
>> server code
>> ldiskfs
>>
>>
>>
>> with LDLM also being explicitly shared, though client & server code is not all shared. Very little to nothing outside of the networking &
>> ldlm layers is shared.
>
> When I first started this (pre 2.3 days) that is how it was explained to
> me. I just took it at face value but when it comes to the separation as I
> don't have a clue. Patrick since you have a better grasp of the
> architecture can you provide details to Neil. Perhaps to documentation on
> this separation if it exist.
James,
you are (or were) correct, but the code has evolved over the years. The lov/osc and
modules were formerly used on the MDS in order to connect to the OSS, and lmv/mdc were
used with the old CMD code to connect to other MDS nodes. However, that ended up with
a lot of code that was in those modules that was only used on the client or server. In
the CMD code it also meant that a lot of complexity existed with making "fake" RPCs to
other MDS nodes that tried to keep the same RPC opcode but were interpreted differently
at the MDS when they were being sent from another MDS.
In the 2.4 release (which included both DNE and the OSD rewrite for ZFS) the lov/osc
server code was copied/split into lod/osp, and this is used for remote connections to
both the OSSs and other MDS nodes. The MDS is still a client of the OSS for precreate
and recovery operations (and soon also statfs), but the code is no longer shared with
the regular clients.
Cheers, Andreas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 873 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/d24975d1/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 7:22 ` Andreas Dilger
@ 2018-06-21 12:28 ` Patrick Farrell
2018-06-24 20:35 ` James Simmons
2018-06-24 23:40 ` NeilBrown
0 siblings, 2 replies; 82+ messages in thread
From: Patrick Farrell @ 2018-06-21 12:28 UTC (permalink / raw)
To: lustre-devel
Ah, thanks Andreas. Perhaps not coincidentally, Lustre 2.4 is the first release I worked on.
Neil, I am really unenamored of the idea of the shared ptlrpc ldlm module being named ptlrpc...
Also, Lustre currently has a bunch of module parameters which are used for configuration. Thoughts on that?
________________________________
From: Andreas Dilger <adilger@dilger.ca>
Sent: Thursday, June 21, 2018 2:22:45 AM
To: James Simmons
Cc: Patrick Farrell; Lustre Development List
Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
On Jun 20, 2018, at 8:57 PM, James Simmons <jsimmons@infradead.org> wrote:
>
>>
>> Huh? Very little of the client is used on the server. The vast majority of the client is llite, vvp, then lov, osc, and mdc. These are
>> not used on the servers at all, the modules won't even be loaded.
>>
>>
>> Servers are definitely *not* clients of one another in the sense of "Lustre clients". They interact with each other, but in a manner very
>> different from client to server.
>>
>> The ldlm and ptlrpc layers are *partly* shared, and of course, the networking from lnet down is. But that's not "the client".
>>
>> The statement that the server layer is a middle layer on top of the client code doesn't make any sense to me. It's like this:
>>
>>
>> vfs
>> client code
>> ptlrpc
>> networking
>> ------ PHYSICAL BOUNDARY --------
>> networking
>> ptlrpc
>> server code
>> ldiskfs
>>
>>
>>
>> with LDLM also being explicitly shared, though client & server code is not all shared. Very little to nothing outside of the networking &
>> ldlm layers is shared.
>
> When I first started this (pre 2.3 days) that is how it was explained to
> me. I just took it at face value but when it comes to the separation as I
> don't have a clue. Patrick since you have a better grasp of the
> architecture can you provide details to Neil. Perhaps to documentation on
> this separation if it exist.
James,
you are (or were) correct, but the code has evolved over the years. The lov/osc and
modules were formerly used on the MDS in order to connect to the OSS, and lmv/mdc were
used with the old CMD code to connect to other MDS nodes. However, that ended up with
a lot of code that was in those modules that was only used on the client or server. In
the CMD code it also meant that a lot of complexity existed with making "fake" RPCs to
other MDS nodes that tried to keep the same RPC opcode but were interpreted differently
at the MDS when they were being sent from another MDS.
In the 2.4 release (which included both DNE and the OSD rewrite for ZFS) the lov/osc
server code was copied/split into lod/osp, and this is used for remote connections to
both the OSSs and other MDS nodes. The MDS is still a client of the OSS for precreate
and recovery operations (and soon also statfs), but the code is no longer shared with
the regular clients.
Cheers, Andreas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180621/79f84741/attachment-0001.html>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid
2018-06-21 4:19 ` NeilBrown
@ 2018-06-22 2:23 ` NeilBrown
2018-06-24 20:33 ` James Simmons
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-22 2:23 UTC (permalink / raw)
To: lustre-devel
On Thu, Jun 21 2018, NeilBrown wrote:
> On Thu, Jun 21 2018, James Simmons wrote:
>
>>> Using spare bits in iattr.ia_valid is not safe as the
>>> bits might get used for some other purpose without
>>> us noticing.
>>> lustre currently used 4 spare bit, but they are all
>>> defined in different places so this isn't immediately
>>> obvious.
>>>
>>> This patch brings all those bit definitions together
>>> and defined a new op_data field (op_xvalid) to hold
>>> the extra validity bits.
>>>
>>> It also replaces sa_valid in struct cl_setattr_io
>>> with sa_avalid and sa_xvalid. Changing the name is
>>> helpful as sa_valid already has another use within
>>> lustre.
>>>
>>> Signed-off-by: NeilBrown <neilb@suse.com>
>>
>> Nak: I see regressions with this patch.
>>
>> [12368.453655] Lustre: DEBUG MARKER: == sanity test 240: race between ldlm
>> enqueue and the connection RPC (no ASSERT) ===================== 21:
>> 16:30 (1529543790)
>> [12368.760832] BUG: Dentry
>> 000000002646a847{i=200004282000008,n=f237.sanity} still in use (1)
>> [unmount of lustre lustre]
>> [12368.773746] WARNING: CPU: 1 PID: 10861 at fs/dcache.c:1514
> ...
>> [12369.385564] Lustre: Unmounted lustre-client
>> [12369.393247] VFS: Busy inodes after unmount of lustre. Self-destruct in
>> 5 seconds. Have a nice day...
>>
>> When I remove this patch things go back to normal. This will not show up
>> if you do a ONLY="240" sh ./sanity.sh. You have to run the sanity.sh in
>> total to make this show up.
>>
>
> (clearly I read my email in the wrong order)
> Very odd. This suggests some sort of life-time-management problem
> with inodes, but the changes shouldn't affect that at all.
> I've gone back over the patch closely and cannot see anything wrong - it
> is really very simple: it just moves flag bits around.
>
> I haven't got as far as 240 yet as changes to kvmalloc have caused
> earlier problem (my VMs don't have much RAM). I'll see if I can
> coax it all the way to 240 and see what happens.
I've now tested all the way to the end and see no new failures.
Specifically test 200 passes.
Have you had it fail more than once? If it was only once, then maybe it
was some random occurrence not related to the patch. We clearly still
want to fix it if we can...
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180622/9e82110f/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c
2018-06-21 1:37 ` James Simmons
@ 2018-06-22 3:19 ` NeilBrown
2018-06-24 20:29 ` James Simmons
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-22 3:19 UTC (permalink / raw)
To: lustre-devel
On Thu, Jun 21 2018, James Simmons wrote:
>> It's good to keep related code together.
>>
>> Signed-off-by: NeilBrown <neilb@suse.com>
>
> Nak. For some reason this corrupts my dmesg output. Its really strange and
> I haven't figured it out yet.
This patch cannot possibly be the cause of any such behavioural change.
It purely moves code from one file to another, it doesn't change the
code at all.
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180622/aaa02fe3/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c
2018-06-22 3:19 ` NeilBrown
@ 2018-06-24 20:29 ` James Simmons
2018-06-24 23:25 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-24 20:29 UTC (permalink / raw)
To: lustre-devel
> On Thu, Jun 21 2018, James Simmons wrote:
>
> >> It's good to keep related code together.
> >>
> >> Signed-off-by: NeilBrown <neilb@suse.com>
> >
> > Nak. For some reason this corrupts my dmesg output. Its really strange and
> > I haven't figured it out yet.
>
> This patch cannot possibly be the cause of any such behavioural change.
> It purely moves code from one file to another, it doesn't change the
> code at all.
In a way you are correct. What this patch does is expose how badly messed
up cfs_print_to_console() is. Originally it was a printk() call but some
newbie changed it incorrectly into a pr_info() for all cases. This is
totally incorrect but it landed anyways :-( For some bizarre reason it
works even with it being completely wrong. I did fix this patch up
and redid the later patches due the changes from the fix up. Is it okay
it I push the newer version of these patches with my changes?
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid
2018-06-22 2:23 ` NeilBrown
@ 2018-06-24 20:33 ` James Simmons
2018-06-24 23:50 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-24 20:33 UTC (permalink / raw)
To: lustre-devel
> On Thu, Jun 21 2018, NeilBrown wrote:
>
> > On Thu, Jun 21 2018, James Simmons wrote:
> >
> >>> Using spare bits in iattr.ia_valid is not safe as the
> >>> bits might get used for some other purpose without
> >>> us noticing.
> >>> lustre currently used 4 spare bit, but they are all
> >>> defined in different places so this isn't immediately
> >>> obvious.
> >>>
> >>> This patch brings all those bit definitions together
> >>> and defined a new op_data field (op_xvalid) to hold
> >>> the extra validity bits.
> >>>
> >>> It also replaces sa_valid in struct cl_setattr_io
> >>> with sa_avalid and sa_xvalid. Changing the name is
> >>> helpful as sa_valid already has another use within
> >>> lustre.
> >>>
> >>> Signed-off-by: NeilBrown <neilb@suse.com>
> >>
> >> Nak: I see regressions with this patch.
> >>
> >> [12368.453655] Lustre: DEBUG MARKER: == sanity test 240: race between ldlm
> >> enqueue and the connection RPC (no ASSERT) ===================== 21:
> >> 16:30 (1529543790)
> >> [12368.760832] BUG: Dentry
> >> 000000002646a847{i=200004282000008,n=f237.sanity} still in use (1)
> >> [unmount of lustre lustre]
> >> [12368.773746] WARNING: CPU: 1 PID: 10861 at fs/dcache.c:1514
> > ...
> >> [12369.385564] Lustre: Unmounted lustre-client
> >> [12369.393247] VFS: Busy inodes after unmount of lustre. Self-destruct in
> >> 5 seconds. Have a nice day...
> >>
> >> When I remove this patch things go back to normal. This will not show up
> >> if you do a ONLY="240" sh ./sanity.sh. You have to run the sanity.sh in
> >> total to make this show up.
> >>
> >
> > (clearly I read my email in the wrong order)
> > Very odd. This suggests some sort of life-time-management problem
> > with inodes, but the changes shouldn't affect that at all.
> > I've gone back over the patch closely and cannot see anything wrong - it
> > is really very simple: it just moves flag bits around.
> >
> > I haven't got as far as 240 yet as changes to kvmalloc have caused
> > earlier problem (my VMs don't have much RAM). I'll see if I can
> > coax it all the way to 240 and see what happens.
>
> I've now tested all the way to the end and see no new failures.
> Specifically test 200 passes.
>
> Have you had it fail more than once? If it was only once, then maybe it
> was some random occurrence not related to the patch. We clearly still
> want to fix it if we can...
Okay I did some more testing and found I get this error with or without
this patch. Its due to a regression from an earlier patch. Can't figure
out where its coming from. I looked at the d_lock usage and it seems
balanced. So all but the first patch for the lustre_compact* header
changes seem okay. BTW they need a rebasing. Also since I have separated
out your libcfs tracefile fixes which I will push your okay with it so
it might be best to break the patches into two series.
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 12:28 ` Patrick Farrell
@ 2018-06-24 20:35 ` James Simmons
2018-06-24 23:40 ` NeilBrown
1 sibling, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-24 20:35 UTC (permalink / raw)
To: lustre-devel
> Ah, thanks Andreas.? Perhaps not coincidentally, Lustre 2.4 is the first release I worked on.
>
> Neil, I am really unenamored of the idea of the shared ptlrpc ldlm module being named ptlrpc...
>
> Also, Lustre currently has a bunch of module parameters which are used for configuration.? Thoughts on that?
Currently ldlm is built into the ptlrpc module for the OpenSFS branch.
Are you suggesting a different approach?
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES
2018-06-21 4:28 ` NeilBrown
@ 2018-06-24 20:37 ` James Simmons
2018-06-24 23:26 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-24 20:37 UTC (permalink / raw)
To: lustre-devel
> As well as CFS_TCD_TYPE_CNT we have TCD_MAX_TYPES which has a larger
> value but a similar meaning. Discard it and just use
> CFS_TCD_TYPE_CNT.
>
> Two places relied on the fact that TCD_MAX_TYPES was larger and so
> there would be NULLs at the end of the array. Change
> them to check the array size properly.
>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
>
> Thanks for testing James!
> I found two problems.
> 1/ I had
> - for (i = 0; cfs_trace_data[i] && i < CFS_TCD_TYPE_CNT; i++) {
> instead of
> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) {
>
> So it could dereference beyond the end of an array. I don't this was
> the problem.
>
> 2/ I hadn't changed
> - for (i = 0; cfs_trace_data[i]; i++) \
> to
> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) \
>
> so if cfs_trace_data[4] was non NULL, bad things could happen.
> I suspect this is what happened to you.
>
> Thanks,
> NeilBrown
Much better :-)
> drivers/staging/lustre/lnet/libcfs/tracefile.c | 8 ++++----
> drivers/staging/lustre/lnet/libcfs/tracefile.h | 5 ++---
> 2 files changed, 6 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.c b/drivers/staging/lustre/lnet/libcfs/tracefile.c
> index cdef67391a72..cc399d580444 100644
> --- a/drivers/staging/lustre/lnet/libcfs/tracefile.c
> +++ b/drivers/staging/lustre/lnet/libcfs/tracefile.c
> @@ -50,7 +50,7 @@
> #include "tracefile.h"
>
> /* XXX move things up to the top, comment */
> -union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS] __cacheline_aligned;
> +union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS] __cacheline_aligned;
>
> char cfs_tracefile[TRACEFILE_NAME_SIZE];
> long long cfs_tracefile_size = CFS_TRACEFILE_SIZE;
> @@ -145,8 +145,8 @@ void cfs_trace_unlock_tcd(struct cfs_trace_cpu_data *tcd, int walking)
> spin_unlock(&tcd->tcd_lock);
> }
>
> -#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
> - for (i = 0; cfs_trace_data[i] && \
> +#define cfs_tcd_for_each_type_lock(tcd, i, cpu) \
> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i] && \
> (tcd = &(*cfs_trace_data[i])[cpu].tcd) && \
> cfs_trace_lock_tcd(tcd, 1); cfs_trace_unlock_tcd(tcd, 1), i++)
>
> @@ -1381,7 +1381,7 @@ static void cfs_trace_cleanup(void)
> cfs_trace_console_buffers[i][j] = NULL;
> }
>
> - for (i = 0; cfs_trace_data[i]; i++) {
> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) {
> kfree(cfs_trace_data[i]);
> cfs_trace_data[i] = NULL;
> }
> diff --git a/drivers/staging/lustre/lnet/libcfs/tracefile.h b/drivers/staging/lustre/lnet/libcfs/tracefile.h
> index 23faecf886c1..87b00fc70b70 100644
> --- a/drivers/staging/lustre/lnet/libcfs/tracefile.h
> +++ b/drivers/staging/lustre/lnet/libcfs/tracefile.h
> @@ -184,11 +184,10 @@ union cfs_trace_data_union {
> char __pad[L1_CACHE_ALIGN(sizeof(struct cfs_trace_cpu_data))];
> };
>
> -#define TCD_MAX_TYPES 8
> -extern union cfs_trace_data_union (*cfs_trace_data[TCD_MAX_TYPES])[NR_CPUS];
> +extern union cfs_trace_data_union (*cfs_trace_data[CFS_TCD_TYPE_CNT])[NR_CPUS];
>
> #define cfs_tcd_for_each(tcd, i, j) \
> - for (i = 0; cfs_trace_data[i]; i++) \
> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) \
> for (j = 0, ((tcd) = &(*cfs_trace_data[i])[j].tcd); \
> j < num_possible_cpus(); \
> j++, (tcd) = &(*cfs_trace_data[i])[j].tcd)
> --
> 2.14.0.rc0.dirty
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 3:00 ` NeilBrown
2018-06-21 3:42 ` Patrick Farrell
@ 2018-06-24 20:39 ` James Simmons
1 sibling, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-24 20:39 UTC (permalink / raw)
To: lustre-devel
> So llite fid fld lmv lov mdc mgc osc and obdclass
> can all be in a "lustre" module.
>
> ldlm and ptlrpc should be in a separate "ptlrpc" module.
> And lnet and the klnds are also separate.
>
> On the server side we would have
> lod mdd mdt mgs ofd osp ost
>
> does that seem approximately right?
The best way to see what modules contain both server and client code is to
search for HAVE_SERVER_SUPPROT in the OpenSFS source tree.
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 2:54 ` NeilBrown
@ 2018-06-24 20:46 ` James Simmons
2018-06-24 23:58 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-24 20:46 UTC (permalink / raw)
To: lustre-devel
> >> Here are a bunch more cleanups that will appear in my lustre-testing
> >> tree shortly.
> >>
> >> The interesting stuff is some enhancements to kbuild to let us
> >> build modules from multiple directories.
> >> I've used that to clean up the build of ptlrpc+ldlm and to
> >> merge libcfs and lnet into the one module - I've left the lnds
> >> separate.
> >
> > That looks pretty good. It doesn't apply to the stable lustre tree but I
> > did try it out on lustre-testing. The module building does seem to
> > function well. Only thing not clear is if do a
> >
> > make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
>
> Why would you do that? SUBDIRS= is for building external modules.
> What are you trying to achieve?
>
> >
> > do lmv.a and the other *.a files get relinked int lustre.ko ?
>
> It isn't "lmv.a", it is ".../lustre/lmv/mod.a".
Sorry I forgot the mod.a naming. I was asking to get a feel of how the
work flow would change. It is common for lustre developers to build
in tree and test without installing anything. The many module approach
might of been a way for people to rebuild a module, unload module and
then reload module quickly to test a change. I don't work this way so
I can't say. I was looking to see if that work flow could still be
done.
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c
2018-06-24 20:29 ` James Simmons
@ 2018-06-24 23:25 ` NeilBrown
2018-06-25 21:52 ` James Simmons
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-24 23:25 UTC (permalink / raw)
To: lustre-devel
On Sun, Jun 24 2018, James Simmons wrote:
>> On Thu, Jun 21 2018, James Simmons wrote:
>>
>> >> It's good to keep related code together.
>> >>
>> >> Signed-off-by: NeilBrown <neilb@suse.com>
>> >
>> > Nak. For some reason this corrupts my dmesg output. Its really strange and
>> > I haven't figured it out yet.
>>
>> This patch cannot possibly be the cause of any such behavioural change.
>> It purely moves code from one file to another, it doesn't change the
>> code at all.
>
> In a way you are correct. What this patch does is expose how badly messed
> up cfs_print_to_console() is. Originally it was a printk() call but some
> newbie changed it incorrectly into a pr_info() for all cases. This is
> totally incorrect but it landed anyways :-( For some bizarre reason it
> works even with it being completely wrong.
I see the problem with cfs_print_to_console(). It chooses a message
type (KERN_EMERG, KERN_ERR, KERN_WARNING, KERN_INFO) then uses pr_info()
to print that, which adds another KERN_INFO.
However vprintk_emit() calls printk_get_level() repeatedly while it
succeeds, so the extra KERNE_INFO is ignored.
So while the code is strange and should use printk(), I don't see
how it could mess up your dmesg output.
> I did fix this patch up
> and redid the later patches due the changes from the fix up. Is it okay
> it I push the newer version of these patches with my changes?
I'm not really sure what you are asking here, but the answer is probably
"yes, it is okay".
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/5ef026fb/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES
2018-06-24 20:37 ` James Simmons
@ 2018-06-24 23:26 ` NeilBrown
0 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-24 23:26 UTC (permalink / raw)
To: lustre-devel
On Sun, Jun 24 2018, James Simmons wrote:
>> As well as CFS_TCD_TYPE_CNT we have TCD_MAX_TYPES which has a larger
>> value but a similar meaning. Discard it and just use
>> CFS_TCD_TYPE_CNT.
>>
>> Two places relied on the fact that TCD_MAX_TYPES was larger and so
>> there would be NULLs at the end of the array. Change
>> them to check the array size properly.
>>
>> Signed-off-by: NeilBrown <neilb@suse.com>
>> ---
>>
>> Thanks for testing James!
>> I found two problems.
>> 1/ I had
>> - for (i = 0; cfs_trace_data[i] && i < CFS_TCD_TYPE_CNT; i++) {
>> instead of
>> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) {
>>
>> So it could dereference beyond the end of an array. I don't this was
>> the problem.
>>
>> 2/ I hadn't changed
>> - for (i = 0; cfs_trace_data[i]; i++) \
>> to
>> + for (i = 0; i < CFS_TCD_TYPE_CNT && cfs_trace_data[i]; i++) \
>>
>> so if cfs_trace_data[4] was non NULL, bad things could happen.
>> I suspect this is what happened to you.
>>
>> Thanks,
>> NeilBrown
>
> Much better :-)
Thanks for the confirmation!
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/499fd3ae/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-21 12:28 ` Patrick Farrell
2018-06-24 20:35 ` James Simmons
@ 2018-06-24 23:40 ` NeilBrown
2018-06-26 1:13 ` James Simmons
1 sibling, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-24 23:40 UTC (permalink / raw)
To: lustre-devel
On Thu, Jun 21 2018, Patrick Farrell wrote:
> Ah, thanks Andreas. Perhaps not coincidentally, Lustre 2.4 is the first release I worked on.
>
> Neil, I am really unenamored of the idea of the shared ptlrpc ldlm module being named ptlrpc...
Can you say why? Is it the choice of name that bothers you, or the
combining of two things into the one kernel module, or something else?
Currently in git://git.hpdd.intel.com/fs/lustre-release.git
(or the git tree that was until recently at the above address),
the module named "ptlrpc" contains ptlrpc code, ldlm code, and target
code. I wonder what "target" means in this context.
>
> Also, Lustre currently has a bunch of module parameters which are used for configuration. Thoughts on that?
Yes, the module parameters are an interesting part of the story.
libcfs has a bunch of module parameters that are symlinked from debugfs.
It seems that user-space largely uses the debugfs links to access them,
so they can become part of the "lnet" module with minimal pain.
Other modules have parameters that I haven't yet looked in to.
There probably will need to be user-space changes to support reduction
in the number of modules.
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/0350635b/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid
2018-06-24 20:33 ` James Simmons
@ 2018-06-24 23:50 ` NeilBrown
0 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-24 23:50 UTC (permalink / raw)
To: lustre-devel
On Sun, Jun 24 2018, James Simmons wrote:
>
> Okay I did some more testing and found I get this error with or without
> this patch. Its due to a regression from an earlier patch. Can't figure
> out where its coming from. I looked at the d_lock usage and it seems
> balanced. So all but the first patch for the lustre_compact* header
> changes seem okay. BTW they need a rebasing. Also since I have separated
> out your libcfs tracefile fixes which I will push your okay with it so
> it might be best to break the patches into two series.
I've changed the problematic lustre_compat patch after a closer look at
the code. I'll post the result shortly.
I don't understand your comment about "break the patches into two
series".
The goal for these patches isn't to make series', it is to apply them to
the "lustre" branch, when they are ready. Some are ready (thanks for
the confirmation), some aren't. The ones that aren't will get revised
and reposted.
I'm okay with you posting any patches that are useful, whether or not
they conflict with anything in my lustre-testing branch. My tendency
will be to apply your patches first, then rebase my outstanding
lustre-testing patches on top of them.
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/f89db3e4/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-24 20:46 ` James Simmons
@ 2018-06-24 23:58 ` NeilBrown
0 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-24 23:58 UTC (permalink / raw)
To: lustre-devel
On Sun, Jun 24 2018, James Simmons wrote:
>> >> Here are a bunch more cleanups that will appear in my lustre-testing
>> >> tree shortly.
>> >>
>> >> The interesting stuff is some enhancements to kbuild to let us
>> >> build modules from multiple directories.
>> >> I've used that to clean up the build of ptlrpc+ldlm and to
>> >> merge libcfs and lnet into the one module - I've left the lnds
>> >> separate.
>> >
>> > That looks pretty good. It doesn't apply to the stable lustre tree but I
>> > did try it out on lustre-testing. The module building does seem to
>> > function well. Only thing not clear is if do a
>> >
>> > make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
>>
>> Why would you do that? SUBDIRS= is for building external modules.
>> What are you trying to achieve?
>>
>> >
>> > do lmv.a and the other *.a files get relinked int lustre.ko ?
>>
>> It isn't "lmv.a", it is ".../lustre/lmv/mod.a".
>
> Sorry I forgot the mod.a naming. I was asking to get a feel of how the
> work flow would change. It is common for lustre developers to build
> in tree and test without installing anything. The many module approach
> might of been a way for people to rebuild a module, unload module and
> then reload module quickly to test a change. I don't work this way so
> I can't say. I was looking to see if that work flow could still be
> done.
This general approach should still work.
You wouldn't do
make SUBDIRS=drivers/staging/lustre/lustre/lmv modules
unless there was a module built in that directory, but you might
make SUBDIRS=drivers/staging/lustre modules
to just build whatever lustre modules there are.
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/d09e4f75/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page()
2018-06-18 2:03 ` NeilBrown
@ 2018-06-25 0:02 ` NeilBrown
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
` (2 more replies)
0 siblings, 3 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-25 0:02 UTC (permalink / raw)
To: lustre-devel
lustre's internal ll_invalidate_page() is behaviourally identical to
generic_error_remove_page().
In the case of lustre it isn't a memory hardware error that requires
the page being invalidated, it is the loss of a lock, which will like
result in the data changing on the server.
In either case, we don't want the page to be accessed any more, so the
same removal is appropriate.
Signed-off-by: NeilBrown <neilb@suse.com>
---
I've replaced
[PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
with 3 patches, this and the next two.
Thanks,
NeilBrown
drivers/staging/lustre/lustre/llite/llite_internal.h | 17 -----------------
drivers/staging/lustre/lustre/llite/vvp_io.c | 2 +-
drivers/staging/lustre/lustre/llite/vvp_page.c | 2 +-
3 files changed, 2 insertions(+), 19 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index c08a6e14b6d7..22dcabf6de0f 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -928,23 +928,6 @@ void policy_from_vma(union ldlm_policy_data *policy, struct vm_area_struct *vma,
struct vm_area_struct *our_vma(struct mm_struct *mm, unsigned long addr,
size_t count);
-static inline void ll_invalidate_page(struct page *vmpage)
-{
- struct address_space *mapping = vmpage->mapping;
- loff_t offset = vmpage->index << PAGE_SHIFT;
-
- LASSERT(PageLocked(vmpage));
- if (!mapping)
- return;
-
- /*
- * truncate_complete_page() calls
- * a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
- */
- ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
- truncate_complete_page(mapping, vmpage);
-}
-
#define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
/* don't need an addref as the sb_info should be holding one */
diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
index e7a4778e02e4..5a67955974ad 100644
--- a/drivers/staging/lustre/lustre/llite/vvp_io.c
+++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
@@ -1098,7 +1098,7 @@ static int vvp_io_fault_start(const struct lu_env *env,
LASSERT(PageLocked(vmpage));
if (OBD_FAIL_CHECK(OBD_FAIL_LLITE_FAULT_TRUNC_RACE))
- ll_invalidate_page(vmpage);
+ generic_error_remove_page(vmpage->mapping, vmpage);
size = i_size_read(inode);
/* Though we have already held a cl_lock upon this page, but
diff --git a/drivers/staging/lustre/lustre/llite/vvp_page.c b/drivers/staging/lustre/lustre/llite/vvp_page.c
index 6eb0565ddc22..dcc4d8faa0cd 100644
--- a/drivers/staging/lustre/lustre/llite/vvp_page.c
+++ b/drivers/staging/lustre/lustre/llite/vvp_page.c
@@ -147,7 +147,7 @@ static void vvp_page_discard(const struct lu_env *env,
if (vpg->vpg_defer_uptodate && !vpg->vpg_ra_used)
ll_ra_stats_inc(vmpage->mapping->host, RA_STAT_DISCARDED);
- ll_invalidate_page(vmpage);
+ generic_error_remove_page(vmpage->mapping, vmpage);
}
static void vvp_page_delete(const struct lu_env *env,
--
2.14.0.rc0.dirty
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/20e31790/attachment-0001.sig>
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages.
2018-06-25 0:02 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() NeilBrown
@ 2018-06-25 0:02 ` NeilBrown
2018-06-25 0:03 ` [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page() NeilBrown
` (2 more replies)
2018-06-25 0:38 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() Oleg Drokin
2018-06-26 0:14 ` James Simmons
2 siblings, 3 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-25 0:02 UTC (permalink / raw)
To: lustre-devel
lustre sometimes uses the internal function truncate_complete_page()
to remove a page of a directory.
Must of what this function does, does not apply to directory pages
as there is no invalidatepage function, and at these times, the
page is not dirty.
The only useful part of the function is delete_from_page_cache(),
so just call that directly.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/staging/lustre/lustre/llite/dir.c | 2 +-
drivers/staging/lustre/lustre/mdc/mdc_request.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
index 688dddf3ca47..52a8ecc2e2e0 100644
--- a/drivers/staging/lustre/lustre/llite/dir.c
+++ b/drivers/staging/lustre/lustre/llite/dir.c
@@ -168,7 +168,7 @@ void ll_release_page(struct inode *inode, struct page *page, bool remove)
if (remove) {
lock_page(page);
if (likely(page->mapping))
- truncate_complete_page(page->mapping, page);
+ delete_from_page_cache(page);
unlock_page(page);
}
put_page(page);
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
index cff31cb0a9ac..827ed0ca0987 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
@@ -927,7 +927,7 @@ static void mdc_release_page(struct page *page, int remove)
if (remove) {
lock_page(page);
if (likely(page->mapping))
- truncate_complete_page(page->mapping, page);
+ delete_from_page_cache(page);
unlock_page(page);
}
put_page(page);
--
2.14.0.rc0.dirty
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/9f5eb849/attachment.sig>
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page()
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
@ 2018-06-25 0:03 ` NeilBrown
2018-06-26 0:19 ` James Simmons
2018-06-25 0:38 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages Oleg Drokin
2018-06-26 0:18 ` James Simmons
2 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-25 0:03 UTC (permalink / raw)
To: lustre-devel
This function is no longer used, so remove it.
Signed-off-by: NeilBrown <neilb@suse.com>
---
.../lustre/lustre/include/lustre_patchless_compat.h | 16 ----------------
1 file changed, 16 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
index 298476ea7557..dca1e3dbd183 100644
--- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
+++ b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
@@ -41,22 +41,6 @@
#include <linux/hash.h>
#include <linux/pagemap.h>
-#define ll_delete_from_page_cache(page) delete_from_page_cache(page)
-
-static inline void
-truncate_complete_page(struct address_space *mapping, struct page *page)
-{
- if (page->mapping != mapping)
- return;
-
- if (PagePrivate(page))
- page->mapping->a_ops->invalidatepage(page, 0, PAGE_SIZE);
-
- cancel_dirty_page(page);
- ClearPageMappedToDisk(page);
- ll_delete_from_page_cache(page);
-}
-
#ifndef ATTR_CTIME_SET
/*
* set ATTR_CTIME_SET to a high value to avoid any risk of collision with other
--
2.14.0.rc0.dirty
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/bb5e4e4c/attachment.sig>
^ permalink raw reply related [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page()
2018-06-25 0:02 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() NeilBrown
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
@ 2018-06-25 0:38 ` Oleg Drokin
2018-06-25 1:57 ` NeilBrown
2018-06-26 0:14 ` James Simmons
2 siblings, 1 reply; 82+ messages in thread
From: Oleg Drokin @ 2018-06-25 0:38 UTC (permalink / raw)
To: lustre-devel
> On Jun 24, 2018, at 8:02 PM, NeilBrown <neilb@suse.com> wrote:
>
>
> lustre's internal ll_invalidate_page() is behaviourally identical to
> generic_error_remove_page().
> In the case of lustre it isn't a memory hardware error that requires
> the page being invalidated, it is the loss of a lock, which will like
> result in the data changing on the server.
> In either case, we don't want the page to be accessed any more, so the
> same removal is appropriate.
>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
>
> I've replaced
> [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
> with 3 patches, this and the next two.
This looks reasonable. Are you running any tests on this?
Acked-by: Oleg Drokin <green@linuxhacker.ru>
>
> Thanks,
> NeilBrown
>
>
> drivers/staging/lustre/lustre/llite/llite_internal.h | 17 -----------------
> drivers/staging/lustre/lustre/llite/vvp_io.c | 2 +-
> drivers/staging/lustre/lustre/llite/vvp_page.c | 2 +-
> 3 files changed, 2 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
> index c08a6e14b6d7..22dcabf6de0f 100644
> --- a/drivers/staging/lustre/lustre/llite/llite_internal.h
> +++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
> @@ -928,23 +928,6 @@ void policy_from_vma(union ldlm_policy_data *policy, struct vm_area_struct *vma,
> struct vm_area_struct *our_vma(struct mm_struct *mm, unsigned long addr,
> size_t count);
>
> -static inline void ll_invalidate_page(struct page *vmpage)
> -{
> - struct address_space *mapping = vmpage->mapping;
> - loff_t offset = vmpage->index << PAGE_SHIFT;
> -
> - LASSERT(PageLocked(vmpage));
> - if (!mapping)
> - return;
> -
> - /*
> - * truncate_complete_page() calls
> - * a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
> - */
> - ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
> - truncate_complete_page(mapping, vmpage);
> -}
> -
> #define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
>
> /* don't need an addref as the sb_info should be holding one */
> diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
> index e7a4778e02e4..5a67955974ad 100644
> --- a/drivers/staging/lustre/lustre/llite/vvp_io.c
> +++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
> @@ -1098,7 +1098,7 @@ static int vvp_io_fault_start(const struct lu_env *env,
> LASSERT(PageLocked(vmpage));
>
> if (OBD_FAIL_CHECK(OBD_FAIL_LLITE_FAULT_TRUNC_RACE))
> - ll_invalidate_page(vmpage);
> + generic_error_remove_page(vmpage->mapping, vmpage);
>
> size = i_size_read(inode);
> /* Though we have already held a cl_lock upon this page, but
> diff --git a/drivers/staging/lustre/lustre/llite/vvp_page.c b/drivers/staging/lustre/lustre/llite/vvp_page.c
> index 6eb0565ddc22..dcc4d8faa0cd 100644
> --- a/drivers/staging/lustre/lustre/llite/vvp_page.c
> +++ b/drivers/staging/lustre/lustre/llite/vvp_page.c
> @@ -147,7 +147,7 @@ static void vvp_page_discard(const struct lu_env *env,
> if (vpg->vpg_defer_uptodate && !vpg->vpg_ra_used)
> ll_ra_stats_inc(vmpage->mapping->host, RA_STAT_DISCARDED);
>
> - ll_invalidate_page(vmpage);
> + generic_error_remove_page(vmpage->mapping, vmpage);
> }
>
> static void vvp_page_delete(const struct lu_env *env,
> --
> 2.14.0.rc0.dirty
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages.
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
2018-06-25 0:03 ` [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page() NeilBrown
@ 2018-06-25 0:38 ` Oleg Drokin
2018-06-26 0:18 ` James Simmons
2 siblings, 0 replies; 82+ messages in thread
From: Oleg Drokin @ 2018-06-25 0:38 UTC (permalink / raw)
To: lustre-devel
> On Jun 24, 2018, at 8:02 PM, NeilBrown <neilb@suse.com> wrote:
>
>
> lustre sometimes uses the internal function truncate_complete_page()
> to remove a page of a directory.
> Must of what this function does, does not apply to directory pages
> as there is no invalidatepage function, and at these times, the
> page is not dirty.
> The only useful part of the function is delete_from_page_cache(),
> so just call that directly.
I guess this is also fine since we indeed do not have dirty directory
pages in Lustre
Acked-by: Oleg Drokin <green@linuxhacker.ru>
>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
> drivers/staging/lustre/lustre/llite/dir.c | 2 +-
> drivers/staging/lustre/lustre/mdc/mdc_request.c | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
> index 688dddf3ca47..52a8ecc2e2e0 100644
> --- a/drivers/staging/lustre/lustre/llite/dir.c
> +++ b/drivers/staging/lustre/lustre/llite/dir.c
> @@ -168,7 +168,7 @@ void ll_release_page(struct inode *inode, struct page *page, bool remove)
> if (remove) {
> lock_page(page);
> if (likely(page->mapping))
> - truncate_complete_page(page->mapping, page);
> + delete_from_page_cache(page);
> unlock_page(page);
> }
> put_page(page);
> diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
> index cff31cb0a9ac..827ed0ca0987 100644
> --- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
> +++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
> @@ -927,7 +927,7 @@ static void mdc_release_page(struct page *page, int remove)
> if (remove) {
> lock_page(page);
> if (likely(page->mapping))
> - truncate_complete_page(page->mapping, page);
> + delete_from_page_cache(page);
> unlock_page(page);
> }
> put_page(page);
> --
> 2.14.0.rc0.dirty
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page()
2018-06-25 0:38 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() Oleg Drokin
@ 2018-06-25 1:57 ` NeilBrown
2018-06-26 0:26 ` James Simmons
0 siblings, 1 reply; 82+ messages in thread
From: NeilBrown @ 2018-06-25 1:57 UTC (permalink / raw)
To: lustre-devel
On Sun, Jun 24 2018, Oleg Drokin wrote:
>> On Jun 24, 2018, at 8:02 PM, NeilBrown <neilb@suse.com> wrote:
>>
>>
>> lustre's internal ll_invalidate_page() is behaviourally identical to
>> generic_error_remove_page().
>> In the case of lustre it isn't a memory hardware error that requires
>> the page being invalidated, it is the loss of a lock, which will like
>> result in the data changing on the server.
>> In either case, we don't want the page to be accessed any more, so the
>> same removal is appropriate.
>>
>> Signed-off-by: NeilBrown <neilb@suse.com>
>> ---
>>
>> I've replaced
>> [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
>> with 3 patches, this and the next two.
>
> This looks reasonable. Are you running any tests on this?
Just the sanity tests on a 4-node vcluster. They haven't caused any
noticeable regressions.
>
> Acked-by: Oleg Drokin <green@linuxhacker.ru>
Thanks,
NeilBrown
>
>>
>> Thanks,
>> NeilBrown
>>
>>
>> drivers/staging/lustre/lustre/llite/llite_internal.h | 17 -----------------
>> drivers/staging/lustre/lustre/llite/vvp_io.c | 2 +-
>> drivers/staging/lustre/lustre/llite/vvp_page.c | 2 +-
>> 3 files changed, 2 insertions(+), 19 deletions(-)
>>
>> diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
>> index c08a6e14b6d7..22dcabf6de0f 100644
>> --- a/drivers/staging/lustre/lustre/llite/llite_internal.h
>> +++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
>> @@ -928,23 +928,6 @@ void policy_from_vma(union ldlm_policy_data *policy, struct vm_area_struct *vma,
>> struct vm_area_struct *our_vma(struct mm_struct *mm, unsigned long addr,
>> size_t count);
>>
>> -static inline void ll_invalidate_page(struct page *vmpage)
>> -{
>> - struct address_space *mapping = vmpage->mapping;
>> - loff_t offset = vmpage->index << PAGE_SHIFT;
>> -
>> - LASSERT(PageLocked(vmpage));
>> - if (!mapping)
>> - return;
>> -
>> - /*
>> - * truncate_complete_page() calls
>> - * a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
>> - */
>> - ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
>> - truncate_complete_page(mapping, vmpage);
>> -}
>> -
>> #define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
>>
>> /* don't need an addref as the sb_info should be holding one */
>> diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
>> index e7a4778e02e4..5a67955974ad 100644
>> --- a/drivers/staging/lustre/lustre/llite/vvp_io.c
>> +++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
>> @@ -1098,7 +1098,7 @@ static int vvp_io_fault_start(const struct lu_env *env,
>> LASSERT(PageLocked(vmpage));
>>
>> if (OBD_FAIL_CHECK(OBD_FAIL_LLITE_FAULT_TRUNC_RACE))
>> - ll_invalidate_page(vmpage);
>> + generic_error_remove_page(vmpage->mapping, vmpage);
>>
>> size = i_size_read(inode);
>> /* Though we have already held a cl_lock upon this page, but
>> diff --git a/drivers/staging/lustre/lustre/llite/vvp_page.c b/drivers/staging/lustre/lustre/llite/vvp_page.c
>> index 6eb0565ddc22..dcc4d8faa0cd 100644
>> --- a/drivers/staging/lustre/lustre/llite/vvp_page.c
>> +++ b/drivers/staging/lustre/lustre/llite/vvp_page.c
>> @@ -147,7 +147,7 @@ static void vvp_page_discard(const struct lu_env *env,
>> if (vpg->vpg_defer_uptodate && !vpg->vpg_ra_used)
>> ll_ra_stats_inc(vmpage->mapping->host, RA_STAT_DISCARDED);
>>
>> - ll_invalidate_page(vmpage);
>> + generic_error_remove_page(vmpage->mapping, vmpage);
>> }
>>
>> static void vvp_page_delete(const struct lu_env *env,
>> --
>> 2.14.0.rc0.dirty
>>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180625/a2a1b8c7/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c
2018-06-24 23:25 ` NeilBrown
@ 2018-06-25 21:52 ` James Simmons
0 siblings, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-25 21:52 UTC (permalink / raw)
To: lustre-devel
> >> On Thu, Jun 21 2018, James Simmons wrote:
> >>
> >> >> It's good to keep related code together.
> >> >>
> >> >> Signed-off-by: NeilBrown <neilb@suse.com>
> >> >
> >> > Nak. For some reason this corrupts my dmesg output. Its really strange and
> >> > I haven't figured it out yet.
> >>
> >> This patch cannot possibly be the cause of any such behavioural change.
> >> It purely moves code from one file to another, it doesn't change the
> >> code at all.
> >
> > In a way you are correct. What this patch does is expose how badly messed
> > up cfs_print_to_console() is. Originally it was a printk() call but some
> > newbie changed it incorrectly into a pr_info() for all cases. This is
> > totally incorrect but it landed anyways :-( For some bizarre reason it
> > works even with it being completely wrong.
>
> I see the problem with cfs_print_to_console(). It chooses a message
> type (KERN_EMERG, KERN_ERR, KERN_WARNING, KERN_INFO) then uses pr_info()
> to print that, which adds another KERN_INFO.
> However vprintk_emit() calls printk_get_level() repeatedly while it
> succeeds, so the extra KERNE_INFO is ignored.
>
> So while the code is strange and should use printk(), I don't see
> how it could mess up your dmesg output.
>
> > I did fix this patch up
> > and redid the later patches due the changes from the fix up. Is it
> > okay it I push the newer version of these patches with my changes?
>
> I'm not really sure what you are asking here, but the answer is probably
> "yes, it is okay".
Pushed the updated patches. Sorry I wasn't clear on what I was asking.
I was weird stuff like Lustre: LNet6: .....
Well in any case the pushed patch set cleans all that mess up.
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page()
2018-06-25 0:02 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() NeilBrown
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
2018-06-25 0:38 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() Oleg Drokin
@ 2018-06-26 0:14 ` James Simmons
2 siblings, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-26 0:14 UTC (permalink / raw)
To: lustre-devel
> lustre's internal ll_invalidate_page() is behaviourally identical to
> generic_error_remove_page().
> In the case of lustre it isn't a memory hardware error that requires
> the page being invalidated, it is the loss of a lock, which will like
> result in the data changing on the server.
> In either case, we don't want the page to be accessed any more, so the
> same removal is appropriate.
Reviewed-by: James Simmons <jsimmons@infradead.org>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
>
> I've replaced
> [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
> with 3 patches, this and the next two.
>
> Thanks,
> NeilBrown
>
>
> drivers/staging/lustre/lustre/llite/llite_internal.h | 17 -----------------
> drivers/staging/lustre/lustre/llite/vvp_io.c | 2 +-
> drivers/staging/lustre/lustre/llite/vvp_page.c | 2 +-
> 3 files changed, 2 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
> index c08a6e14b6d7..22dcabf6de0f 100644
> --- a/drivers/staging/lustre/lustre/llite/llite_internal.h
> +++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
> @@ -928,23 +928,6 @@ void policy_from_vma(union ldlm_policy_data *policy, struct vm_area_struct *vma,
> struct vm_area_struct *our_vma(struct mm_struct *mm, unsigned long addr,
> size_t count);
>
> -static inline void ll_invalidate_page(struct page *vmpage)
> -{
> - struct address_space *mapping = vmpage->mapping;
> - loff_t offset = vmpage->index << PAGE_SHIFT;
> -
> - LASSERT(PageLocked(vmpage));
> - if (!mapping)
> - return;
> -
> - /*
> - * truncate_complete_page() calls
> - * a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
> - */
> - ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
> - truncate_complete_page(mapping, vmpage);
> -}
> -
> #define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
>
> /* don't need an addref as the sb_info should be holding one */
> diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
> index e7a4778e02e4..5a67955974ad 100644
> --- a/drivers/staging/lustre/lustre/llite/vvp_io.c
> +++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
> @@ -1098,7 +1098,7 @@ static int vvp_io_fault_start(const struct lu_env *env,
> LASSERT(PageLocked(vmpage));
>
> if (OBD_FAIL_CHECK(OBD_FAIL_LLITE_FAULT_TRUNC_RACE))
> - ll_invalidate_page(vmpage);
> + generic_error_remove_page(vmpage->mapping, vmpage);
>
> size = i_size_read(inode);
> /* Though we have already held a cl_lock upon this page, but
> diff --git a/drivers/staging/lustre/lustre/llite/vvp_page.c b/drivers/staging/lustre/lustre/llite/vvp_page.c
> index 6eb0565ddc22..dcc4d8faa0cd 100644
> --- a/drivers/staging/lustre/lustre/llite/vvp_page.c
> +++ b/drivers/staging/lustre/lustre/llite/vvp_page.c
> @@ -147,7 +147,7 @@ static void vvp_page_discard(const struct lu_env *env,
> if (vpg->vpg_defer_uptodate && !vpg->vpg_ra_used)
> ll_ra_stats_inc(vmpage->mapping->host, RA_STAT_DISCARDED);
>
> - ll_invalidate_page(vmpage);
> + generic_error_remove_page(vmpage->mapping, vmpage);
> }
>
> static void vvp_page_delete(const struct lu_env *env,
> --
> 2.14.0.rc0.dirty
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages.
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
2018-06-25 0:03 ` [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page() NeilBrown
2018-06-25 0:38 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages Oleg Drokin
@ 2018-06-26 0:18 ` James Simmons
2 siblings, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-26 0:18 UTC (permalink / raw)
To: lustre-devel
> lustre sometimes uses the internal function truncate_complete_page()
> to remove a page of a directory.
> Must of what this function does, does not apply to directory pages
> as there is no invalidatepage function, and at these times, the
> page is not dirty.
> The only useful part of the function is delete_from_page_cache(),
> so just call that directly.
>
Reviewed-by: James Simmons <jsimmons@infradead.org>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
> drivers/staging/lustre/lustre/llite/dir.c | 2 +-
> drivers/staging/lustre/lustre/mdc/mdc_request.c | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
> index 688dddf3ca47..52a8ecc2e2e0 100644
> --- a/drivers/staging/lustre/lustre/llite/dir.c
> +++ b/drivers/staging/lustre/lustre/llite/dir.c
> @@ -168,7 +168,7 @@ void ll_release_page(struct inode *inode, struct page *page, bool remove)
> if (remove) {
> lock_page(page);
> if (likely(page->mapping))
> - truncate_complete_page(page->mapping, page);
> + delete_from_page_cache(page);
> unlock_page(page);
> }
> put_page(page);
> diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
> index cff31cb0a9ac..827ed0ca0987 100644
> --- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
> +++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
> @@ -927,7 +927,7 @@ static void mdc_release_page(struct page *page, int remove)
> if (remove) {
> lock_page(page);
> if (likely(page->mapping))
> - truncate_complete_page(page->mapping, page);
> + delete_from_page_cache(page);
> unlock_page(page);
> }
> put_page(page);
> --
> 2.14.0.rc0.dirty
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page()
2018-06-25 0:03 ` [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page() NeilBrown
@ 2018-06-26 0:19 ` James Simmons
0 siblings, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-26 0:19 UTC (permalink / raw)
To: lustre-devel
> This function is no longer used, so remove it.
Reviewed-by: James Simmons <jsimmons@infradead.org>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
> .../lustre/lustre/include/lustre_patchless_compat.h | 16 ----------------
> 1 file changed, 16 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> index 298476ea7557..dca1e3dbd183 100644
> --- a/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> +++ b/drivers/staging/lustre/lustre/include/lustre_patchless_compat.h
> @@ -41,22 +41,6 @@
> #include <linux/hash.h>
> #include <linux/pagemap.h>
>
> -#define ll_delete_from_page_cache(page) delete_from_page_cache(page)
> -
> -static inline void
> -truncate_complete_page(struct address_space *mapping, struct page *page)
> -{
> - if (page->mapping != mapping)
> - return;
> -
> - if (PagePrivate(page))
> - page->mapping->a_ops->invalidatepage(page, 0, PAGE_SIZE);
> -
> - cancel_dirty_page(page);
> - ClearPageMappedToDisk(page);
> - ll_delete_from_page_cache(page);
> -}
> -
> #ifndef ATTR_CTIME_SET
> /*
> * set ATTR_CTIME_SET to a high value to avoid any risk of collision with other
> --
> 2.14.0.rc0.dirty
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page()
2018-06-25 1:57 ` NeilBrown
@ 2018-06-26 0:26 ` James Simmons
0 siblings, 0 replies; 82+ messages in thread
From: James Simmons @ 2018-06-26 0:26 UTC (permalink / raw)
To: lustre-devel
> >> On Jun 24, 2018, at 8:02 PM, NeilBrown <neilb@suse.com> wrote:
> >>
> >>
> >> lustre's internal ll_invalidate_page() is behaviourally identical to
> >> generic_error_remove_page().
> >> In the case of lustre it isn't a memory hardware error that requires
> >> the page being invalidated, it is the loss of a lock, which will like
> >> result in the data changing on the server.
> >> In either case, we don't want the page to be accessed any more, so the
> >> same removal is appropriate.
> >>
> >> Signed-off-by: NeilBrown <neilb@suse.com>
> >> ---
> >>
> >> I've replaced
> >> [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page
> >> with 3 patches, this and the next two.
> >
> > This looks reasonable. Are you running any tests on this?
>
> Just the sanity tests on a 4-node vcluster. They haven't caused any
> noticeable regressions.
>
I also have tested this patch set using a single client node with
a stand alone MGS server, 2 MDS servers with 2 MDTs each, and a
single OSS server with 2 OST.
> >
> > Acked-by: Oleg Drokin <green@linuxhacker.ru>
>
> Thanks,
> NeilBrown
>
> >
> >>
> >> Thanks,
> >> NeilBrown
> >>
> >>
> >> drivers/staging/lustre/lustre/llite/llite_internal.h | 17 -----------------
> >> drivers/staging/lustre/lustre/llite/vvp_io.c | 2 +-
> >> drivers/staging/lustre/lustre/llite/vvp_page.c | 2 +-
> >> 3 files changed, 2 insertions(+), 19 deletions(-)
> >>
> >> diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
> >> index c08a6e14b6d7..22dcabf6de0f 100644
> >> --- a/drivers/staging/lustre/lustre/llite/llite_internal.h
> >> +++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
> >> @@ -928,23 +928,6 @@ void policy_from_vma(union ldlm_policy_data *policy, struct vm_area_struct *vma,
> >> struct vm_area_struct *our_vma(struct mm_struct *mm, unsigned long addr,
> >> size_t count);
> >>
> >> -static inline void ll_invalidate_page(struct page *vmpage)
> >> -{
> >> - struct address_space *mapping = vmpage->mapping;
> >> - loff_t offset = vmpage->index << PAGE_SHIFT;
> >> -
> >> - LASSERT(PageLocked(vmpage));
> >> - if (!mapping)
> >> - return;
> >> -
> >> - /*
> >> - * truncate_complete_page() calls
> >> - * a_ops->invalidatepage()->cl_page_delete()->vvp_page_delete().
> >> - */
> >> - ll_teardown_mmaps(mapping, offset, offset + PAGE_SIZE);
> >> - truncate_complete_page(mapping, vmpage);
> >> -}
> >> -
> >> #define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi)
> >>
> >> /* don't need an addref as the sb_info should be holding one */
> >> diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
> >> index e7a4778e02e4..5a67955974ad 100644
> >> --- a/drivers/staging/lustre/lustre/llite/vvp_io.c
> >> +++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
> >> @@ -1098,7 +1098,7 @@ static int vvp_io_fault_start(const struct lu_env *env,
> >> LASSERT(PageLocked(vmpage));
> >>
> >> if (OBD_FAIL_CHECK(OBD_FAIL_LLITE_FAULT_TRUNC_RACE))
> >> - ll_invalidate_page(vmpage);
> >> + generic_error_remove_page(vmpage->mapping, vmpage);
> >>
> >> size = i_size_read(inode);
> >> /* Though we have already held a cl_lock upon this page, but
> >> diff --git a/drivers/staging/lustre/lustre/llite/vvp_page.c b/drivers/staging/lustre/lustre/llite/vvp_page.c
> >> index 6eb0565ddc22..dcc4d8faa0cd 100644
> >> --- a/drivers/staging/lustre/lustre/llite/vvp_page.c
> >> +++ b/drivers/staging/lustre/lustre/llite/vvp_page.c
> >> @@ -147,7 +147,7 @@ static void vvp_page_discard(const struct lu_env *env,
> >> if (vpg->vpg_defer_uptodate && !vpg->vpg_ra_used)
> >> ll_ra_stats_inc(vmpage->mapping->host, RA_STAT_DISCARDED);
> >>
> >> - ll_invalidate_page(vmpage);
> >> + generic_error_remove_page(vmpage->mapping, vmpage);
> >> }
> >>
> >> static void vvp_page_delete(const struct lu_env *env,
> >> --
> >> 2.14.0.rc0.dirty
> >>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-24 23:40 ` NeilBrown
@ 2018-06-26 1:13 ` James Simmons
2018-06-26 13:51 ` Patrick Farrell
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-06-26 1:13 UTC (permalink / raw)
To: lustre-devel
> > Ah, thanks Andreas. Perhaps not coincidentally, Lustre 2.4 is the first release I worked on.
> >
> > Neil, I am really unenamored of the idea of the shared ptlrpc ldlm module being named ptlrpc...
>
> Can you say why? Is it the choice of name that bothers you, or the
> combining of two things into the one kernel module, or something else?
> Currently in git://git.hpdd.intel.com/fs/lustre-release.git
> (or the git tree that was until recently at the above address),
> the module named "ptlrpc" contains ptlrpc code, ldlm code, and target
> code. I wonder what "target" means in this context.
target is the server code built into ptlrpc which is for the "Unified
Target". Its the shared code that is used by both the MDS and OSS code.
Andreas or Oleg can add more details than this. BTW also the nodemap
code also get built into ptlrpc module as well for the server side.
> > Also, Lustre currently has a bunch of module parameters which are used for configuration. Thoughts on that?
>
> Yes, the module parameters are an interesting part of the story.
> libcfs has a bunch of module parameters that are symlinked from debugfs.
> It seems that user-space largely uses the debugfs links to access them,
> so they can become part of the "lnet" module with minimal pain.
> Other modules have parameters that I haven't yet looked in to.
>
> There probably will need to be user-space changes to support reduction
> in the number of modules.
The test suite will need some changes since it loads the modules. We
need to handle the case if its built into the kernel as well.
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-26 1:13 ` James Simmons
@ 2018-06-26 13:51 ` Patrick Farrell
2018-06-27 3:08 ` NeilBrown
0 siblings, 1 reply; 82+ messages in thread
From: Patrick Farrell @ 2018-06-26 13:51 UTC (permalink / raw)
To: lustre-devel
Ah, sorry, lost your mail in the shuffle, Neil.
The name, mostly. ptlrpc is one subdirectory and one subsystem, so I don?t want to use it for a module that explicitly includes all of several. I?m not sure of a better name immediately - is this module intended to be shared between client and server?
________________________________
From: James Simmons <jsimmons@infradead.org>
Sent: Monday, June 25, 2018 8:13:22 PM
To: NeilBrown
Cc: Patrick Farrell; Andreas Dilger; Oleg Drokin; Lustre Development List
Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
> > Ah, thanks Andreas. Perhaps not coincidentally, Lustre 2.4 is the first release I worked on.
> >
> > Neil, I am really unenamored of the idea of the shared ptlrpc ldlm module being named ptlrpc...
>
> Can you say why? Is it the choice of name that bothers you, or the
> combining of two things into the one kernel module, or something else?
> Currently in git://git.hpdd.intel.com/fs/lustre-release.git
> (or the git tree that was until recently at the above address),
> the module named "ptlrpc" contains ptlrpc code, ldlm code, and target
> code. I wonder what "target" means in this context.
target is the server code built into ptlrpc which is for the "Unified
Target". Its the shared code that is used by both the MDS and OSS code.
Andreas or Oleg can add more details than this. BTW also the nodemap
code also get built into ptlrpc module as well for the server side.
> > Also, Lustre currently has a bunch of module parameters which are used for configuration. Thoughts on that?
>
> Yes, the module parameters are an interesting part of the story.
> libcfs has a bunch of module parameters that are symlinked from debugfs.
> It seems that user-space largely uses the debugfs links to access them,
> so they can become part of the "lnet" module with minimal pain.
> Other modules have parameters that I haven't yet looked in to.
>
> There probably will need to be user-space changes to support reduction
> in the number of modules.
The test suite will need some changes since it loads the modules. We
need to handle the case if its built into the kernel as well.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180626/bc8c6124/attachment.html>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-26 13:51 ` Patrick Farrell
@ 2018-06-27 3:08 ` NeilBrown
2018-06-27 4:00 ` Cory Spitz
2018-06-27 11:01 ` Andreas Dilger
0 siblings, 2 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-27 3:08 UTC (permalink / raw)
To: lustre-devel
On Tue, Jun 26 2018, Patrick Farrell wrote:
> Ah, sorry, lost your mail in the shuffle, Neil.
>
> The name, mostly. ptlrpc is one subdirectory and one subsystem, so I don?t want to use it for a module that explicitly includes all of several. I?m not sure of a better name immediately - is this module intended to be shared between client and server?
>
Thanks for clarifying.
Note that I'm still feeling my way around here so "intended" can at most
be a soft intention, open to change. But yes, I expect this module
would ultimately be shared with the server.
I think things should only be in different modules if it might sometimes
make sense to use one without the other.
So everything that is always used for a lustre client mount, and is only
used for a lustre client mount, should be in a module called "lustre".
And everything always and only used for a lustre server should be in a
module called (something like) "lustre-server".
It appears to me that ptlrpc, ldlm, and obdclass are always and only
used together, but can be used for client or server. So I think they
should be together in a module.
I'm tempted to add lnet to that list. There has been mention that cray
has some other functionality that uses lnet - if that doesn't use
ptlrpc, then maybe that is a case for keeping lnet separate. However I
would rather see it as a potential case to separate lnet from the others
at a later date if that functionality from cray ever goes into mainline.
The klnds modules can presumably be used independently(?) so they can
sensibly remain as separate modules, though I'm beginning to wonder if
lnet can actually function without socklnd, so I wonder if that should
be permanently part of the lnet module (with NFS, the xprt_sock code is
a permanent part of "sunrpc", while the xprt_rdma (aka rpcrdma) code
can be a separate module).
In the interests of a concrete strawman, what objections would I get if
I suggested that ptlrpc, ldlm, obdclass, lnet, and socklnd were all included in
the one module named "lnet" ??
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180627/02d56082/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-27 3:08 ` NeilBrown
@ 2018-06-27 4:00 ` Cory Spitz
2018-06-28 1:26 ` NeilBrown
2018-06-27 11:01 ` Andreas Dilger
1 sibling, 1 reply; 82+ messages in thread
From: Cory Spitz @ 2018-06-27 4:00 UTC (permalink / raw)
To: lustre-devel
Hello, Neil.
Cray does indeed have another user of LNet. While it is now GPL, I can't say if it will ever be in mainline. However, even without it there is a case for LNet to live on its own as Andreas pointed out recently with http://lists.lustre.org/pipermail/lustre-devel-lustre.org/2018-June/007098.html.
But, if you want to group everything together now, we can hopefully tease it apart again in the future. If so, I think that it would be misleading to call a module lnet if it contained distributed locking functionality (ldlm). How about lustre-common, lustre-core, or some such?
Thanks,
-Cory
--
?On 6/26/18, 10:09 PM, "lustre-devel on behalf of NeilBrown" <lustre-devel-bounces at lists.lustre.org on behalf of neilb@suse.com> wrote:
On Tue, Jun 26 2018, Patrick Farrell wrote:
> Ah, sorry, lost your mail in the shuffle, Neil.
>
> The name, mostly. ptlrpc is one subdirectory and one subsystem, so I don?t want to use it for a module that explicitly includes all of several. I?m not sure of a better name immediately - is this module intended to be shared between client and server?
>
Thanks for clarifying.
Note that I'm still feeling my way around here so "intended" can at most
be a soft intention, open to change. But yes, I expect this module
would ultimately be shared with the server.
I think things should only be in different modules if it might sometimes
make sense to use one without the other.
So everything that is always used for a lustre client mount, and is only
used for a lustre client mount, should be in a module called "lustre".
And everything always and only used for a lustre server should be in a
module called (something like) "lustre-server".
It appears to me that ptlrpc, ldlm, and obdclass are always and only
used together, but can be used for client or server. So I think they
should be together in a module.
I'm tempted to add lnet to that list. There has been mention that cray
has some other functionality that uses lnet - if that doesn't use
ptlrpc, then maybe that is a case for keeping lnet separate. However I
would rather see it as a potential case to separate lnet from the others
at a later date if that functionality from cray ever goes into mainline.
The klnds modules can presumably be used independently(?) so they can
sensibly remain as separate modules, though I'm beginning to wonder if
lnet can actually function without socklnd, so I wonder if that should
be permanently part of the lnet module (with NFS, the xprt_sock code is
a permanent part of "sunrpc", while the xprt_rdma (aka rpcrdma) code
can be a separate module).
In the interests of a concrete strawman, what objections would I get if
I suggested that ptlrpc, ldlm, obdclass, lnet, and socklnd were all included in
the one module named "lnet" ??
Thanks,
NeilBrown
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-27 3:08 ` NeilBrown
2018-06-27 4:00 ` Cory Spitz
@ 2018-06-27 11:01 ` Andreas Dilger
2018-06-27 12:06 ` Patrick Farrell
2018-06-28 1:39 ` NeilBrown
1 sibling, 2 replies; 82+ messages in thread
From: Andreas Dilger @ 2018-06-27 11:01 UTC (permalink / raw)
To: lustre-devel
On Jun 27, 2018, at 05:08, NeilBrown <neilb@suse.com> wrote:
>
> On Tue, Jun 26 2018, Patrick Farrell wrote:
>
>> Ah, sorry, lost your mail in the shuffle, Neil.
>>
>> The name, mostly. ptlrpc is one subdirectory and one subsystem, so I don?t want to use it for a module that explicitly includes all of several. I?m not sure of a better name immediately - is this module intended to be shared between client and server?
>>
>
> Thanks for clarifying.
> Note that I'm still feeling my way around here so "intended" can at most
> be a soft intention, open to change. But yes, I expect this module
> would ultimately be shared with the server.
>
> I think things should only be in different modules if it might sometimes
> make sense to use one without the other.
>
> So everything that is always used for a lustre client mount, and is only
> used for a lustre client mount, should be in a module called "lustre".
> And everything always and only used for a lustre server should be in a
> module called (something like) "lustre-server".
>
> It appears to me that ptlrpc, ldlm, and obdclass are always and only
> used together, but can be used for client or server. So I think they
> should be together in a module.
>
> I'm tempted to add lnet to that list. There has been mention that Cray
> has some other functionality that uses lnet - if that doesn't use
> ptlrpc, then maybe that is a case for keeping lnet separate. However I
> would rather see it as a potential case to separate lnet from the others
> at a later date if that functionality from cray ever goes into mainline.
>
> The klnds modules can presumably be used independently(?) so they can
> sensibly remain as separate modules, though I'm beginning to wonder if
> lnet can actually function without socklnd, so I wonder if that should
> be permanently part of the lnet module (with NFS, the xprt_sock code is
> a permanent part of "sunrpc", while the xprt_rdma (aka rpcrdma) code
> can be a separate module).
Definitely LNet can run without ksocklnd, and it does so on clients that
only have IB connections. Lots of HPC systems have client nodes that do
not have Ethernet interfaces, since it is a cabling/networking nightmare
to add a few thousand additional Ethernet cables/ports in addition to the
IB cables/ports, and it is always possible to run IPoIB for TCP traffic.
> In the interests of a concrete strawman, what objections would I get if
> I suggested that ptlrpc, ldlm, obdclass, lnet, and socklnd were all included in
> the one module named "lnet" ??
This would break all of the module parameters for ptlrpc. There are only
a few obdclass module parameters that are rarely used, so I'm less worried
about those ones.
Given that lnet is a separate thing, I'd prefer to keep "ptlrpc" and "lnet"
as separate modules, then "lustre" for the remainder of the client code.
The "ptlrpc" module should also include the "fld", "fid", "quota", "mgc"
modules, since these are shared with the server as well.
Server modules:
[adilger at mookie ~]$ lsmod | grep ^obdclass
obdclass 1731964 83 mdd,lod,mdt,osp,ofd,lfsck,ost,mgs,mgc,osd_ldiskfs,fid,fld,lquota,ptlrpc
Client modules:
[adilger at twoshoes ~]$ lsmod | grep ^obdclass
obdclass 1568040 86 osc,mgc,lustre,lov,mdc,fid,lmv,fld,ptlrpc
James, if we are going to move in the direction of having a separate
"lustre-server" module, it would make sense to land a patch to master
now that adds both a module alias, as well as a filesystem type alias
for lustre-server, so that we can start moving systems over to using
"lustre-server" as the filesystem type in /etc/fstab or HA mounting
scripts, and it will auto-load the module at the same time. This might
also allow the Lustre mount code to be simplified, since I think it has
some messy code to have the same code, but distinguish between client
and server mounts.
Cheers, Andreas
---
Andreas Dilger
Principal Lustre Architect
Whamcloud
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 235 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180627/91d93896/attachment-0001.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-27 11:01 ` Andreas Dilger
@ 2018-06-27 12:06 ` Patrick Farrell
2018-06-28 1:59 ` NeilBrown
2018-06-28 1:39 ` NeilBrown
1 sibling, 1 reply; 82+ messages in thread
From: Patrick Farrell @ 2018-06-27 12:06 UTC (permalink / raw)
To: lustre-devel
Neil,
We do indeed have such functionality (it?s called DVS and it?s basically a high speed file system projection framework, ala NFS but faster), so the ability to build lnet separately is valuable to us. While it is being open sourced under the GPL, I don?t think there?s any intention to try to upstream it. The current code isn?t even usable off of Cray systems as it depends on info from user space (that is provided, in the end, from Cray proprietary hardware) to keep its connection/routing tables up to date. That?s supposedly in the pipeline to get fixed, but it?s still pretty far from generally usable.
But we?d still really appreciate it if lnet stayed separate. Don?t know if that?s enough for you - I know sometimes *small* stuff is done for out of tree users. Hopefully this meets that standard.
- Patrick
________________________________
From: Andreas Dilger <adilger@whamcloud.com>
Sent: Wednesday, June 27, 2018 6:01:27 AM
To: NeilBrown; James Simmons
Cc: Patrick Farrell; Oleg Drokin; Lustre Development List
Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
On Jun 27, 2018, at 05:08, NeilBrown <neilb@suse.com> wrote:
>
> On Tue, Jun 26 2018, Patrick Farrell wrote:
>
>> Ah, sorry, lost your mail in the shuffle, Neil.
>>
>> The name, mostly. ptlrpc is one subdirectory and one subsystem, so I don?t want to use it for a module that explicitly includes all of several. I?m not sure of a better name immediately - is this module intended to be shared between client and server?
>>
>
> Thanks for clarifying.
> Note that I'm still feeling my way around here so "intended" can at most
> be a soft intention, open to change. But yes, I expect this module
> would ultimately be shared with the server.
>
> I think things should only be in different modules if it might sometimes
> make sense to use one without the other.
>
> So everything that is always used for a lustre client mount, and is only
> used for a lustre client mount, should be in a module called "lustre".
> And everything always and only used for a lustre server should be in a
> module called (something like) "lustre-server".
>
> It appears to me that ptlrpc, ldlm, and obdclass are always and only
> used together, but can be used for client or server. So I think they
> should be together in a module.
>
> I'm tempted to add lnet to that list. There has been mention that Cray
> has some other functionality that uses lnet - if that doesn't use
> ptlrpc, then maybe that is a case for keeping lnet separate. However I
> would rather see it as a potential case to separate lnet from the others
> at a later date if that functionality from cray ever goes into mainline.
>
> The klnds modules can presumably be used independently(?) so they can
> sensibly remain as separate modules, though I'm beginning to wonder if
> lnet can actually function without socklnd, so I wonder if that should
> be permanently part of the lnet module (with NFS, the xprt_sock code is
> a permanent part of "sunrpc", while the xprt_rdma (aka rpcrdma) code
> can be a separate module).
Definitely LNet can run without ksocklnd, and it does so on clients that
only have IB connections. Lots of HPC systems have client nodes that do
not have Ethernet interfaces, since it is a cabling/networking nightmare
to add a few thousand additional Ethernet cables/ports in addition to the
IB cables/ports, and it is always possible to run IPoIB for TCP traffic.
> In the interests of a concrete strawman, what objections would I get if
> I suggested that ptlrpc, ldlm, obdclass, lnet, and socklnd were all included in
> the one module named "lnet" ??
This would break all of the module parameters for ptlrpc. There are only
a few obdclass module parameters that are rarely used, so I'm less worried
about those ones.
Given that lnet is a separate thing, I'd prefer to keep "ptlrpc" and "lnet"
as separate modules, then "lustre" for the remainder of the client code.
The "ptlrpc" module should also include the "fld", "fid", "quota", "mgc"
modules, since these are shared with the server as well.
Server modules:
[adilger at mookie ~]$ lsmod | grep ^obdclass
obdclass 1731964 83 mdd,lod,mdt,osp,ofd,lfsck,ost,mgs,mgc,osd_ldiskfs,fid,fld,lquota,ptlrpc
Client modules:
[adilger at twoshoes ~]$ lsmod | grep ^obdclass
obdclass 1568040 86 osc,mgc,lustre,lov,mdc,fid,lmv,fld,ptlrpc
James, if we are going to move in the direction of having a separate
"lustre-server" module, it would make sense to land a patch to master
now that adds both a module alias, as well as a filesystem type alias
for lustre-server, so that we can start moving systems over to using
"lustre-server" as the filesystem type in /etc/fstab or HA mounting
scripts, and it will auto-load the module at the same time. This might
also allow the Lustre mount code to be simplified, since I think it has
some messy code to have the same code, but distinguish between client
and server mounts.
Cheers, Andreas
---
Andreas Dilger
Principal Lustre Architect
Whamcloud
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180627/b0c53d64/attachment.html>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-27 4:00 ` Cory Spitz
@ 2018-06-28 1:26 ` NeilBrown
0 siblings, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-28 1:26 UTC (permalink / raw)
To: lustre-devel
On Wed, Jun 27 2018, Cory Spitz wrote:
> Hello, Neil.
>
> Cray does indeed have another user of LNet. While it is now GPL, I can't say if it will ever be in mainline. However, even without it there is a case for LNet to live on its own as Andreas pointed out recently with http://lists.lustre.org/pipermail/lustre-devel-lustre.org/2018-June/007098.html.
>
> But, if you want to group everything together now, we can hopefully tease it apart again in the future. If so, I think that it would be misleading to call a module lnet if it contained distributed locking functionality (ldlm). How about lustre-common, lustre-core, or some such?
>
Do you have a link to this open-source thing that uses LNet? Does it
have a name? Is it the Zest thing that Andreas' email linked to? Or
maybe it's Cray-DVS?? A link to a git tree would be very helpful.
(I'm not committed to any name before I understand what the encapsulated
functionality would be)
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180628/2a55161c/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-27 11:01 ` Andreas Dilger
2018-06-27 12:06 ` Patrick Farrell
@ 2018-06-28 1:39 ` NeilBrown
1 sibling, 0 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-28 1:39 UTC (permalink / raw)
To: lustre-devel
On Wed, Jun 27 2018, Andreas Dilger wrote:
> On Jun 27, 2018, at 05:08, NeilBrown <neilb@suse.com> wrote:
>>
>> On Tue, Jun 26 2018, Patrick Farrell wrote:
>>
>>> Ah, sorry, lost your mail in the shuffle, Neil.
>>>
>>> The name, mostly. ptlrpc is one subdirectory and one subsystem, so I don?t want to use it for a module that explicitly includes all of several. I?m not sure of a better name immediately - is this module intended to be shared between client and server?
>>>
>>
>> Thanks for clarifying.
>> Note that I'm still feeling my way around here so "intended" can at most
>> be a soft intention, open to change. But yes, I expect this module
>> would ultimately be shared with the server.
>>
>> I think things should only be in different modules if it might sometimes
>> make sense to use one without the other.
>>
>> So everything that is always used for a lustre client mount, and is only
>> used for a lustre client mount, should be in a module called "lustre".
>> And everything always and only used for a lustre server should be in a
>> module called (something like) "lustre-server".
>>
>> It appears to me that ptlrpc, ldlm, and obdclass are always and only
>> used together, but can be used for client or server. So I think they
>> should be together in a module.
>>
>> I'm tempted to add lnet to that list. There has been mention that Cray
>> has some other functionality that uses lnet - if that doesn't use
>> ptlrpc, then maybe that is a case for keeping lnet separate. However I
>> would rather see it as a potential case to separate lnet from the others
>> at a later date if that functionality from cray ever goes into mainline.
>>
>> The klnds modules can presumably be used independently(?) so they can
>> sensibly remain as separate modules, though I'm beginning to wonder if
>> lnet can actually function without socklnd, so I wonder if that should
>> be permanently part of the lnet module (with NFS, the xprt_sock code is
>> a permanent part of "sunrpc", while the xprt_rdma (aka rpcrdma) code
>> can be a separate module).
>
> Definitely LNet can run without ksocklnd, and it does so on clients that
> only have IB connections. Lots of HPC systems have client nodes that do
> not have Ethernet interfaces, since it is a cabling/networking nightmare
> to add a few thousand additional Ethernet cables/ports in addition to the
> IB cables/ports, and it is always possible to run IPoIB for TCP traffic.
Thanks for the clarification.
This came up because I'd been looking at lnet/lib-socket.c and wondering
why that code wasn't in socklnd. If found that it was also used by
lnet/acceptor which seems to always run a thread that listens on a TCP
port (I think - I skim lots of code while trying to figure out the
big-picture structure).
If there is always a socket acceptor, I thought you might always need a
socket back-end (lnd).
Should the 'acceptor' thing really be part of socklnd?
>
>> In the interests of a concrete strawman, what objections would I get if
>> I suggested that ptlrpc, ldlm, obdclass, lnet, and socklnd were all included in
>> the one module named "lnet" ??
>
> This would break all of the module parameters for ptlrpc. There are only
> a few obdclass module parameters that are rarely used, so I'm less worried
> about those ones.
>
> Given that lnet is a separate thing, I'd prefer to keep "ptlrpc" and "lnet"
> as separate modules, then "lustre" for the remainder of the client
> code.
This does seem to be the idea that will be most widely accepted.
The module parameters seem to be the main practical issue with merging
modules. I wouldn't want that to force us to keep things separate that
should be together, but nor do I want to introduce unnecessary
breakage.
>
> The "ptlrpc" module should also include the "fld", "fid", "quota", "mgc"
> modules, since these are shared with the server as well.
>
> Server modules:
> [adilger at mookie ~]$ lsmod | grep ^obdclass
> obdclass 1731964 83 mdd,lod,mdt,osp,ofd,lfsck,ost,mgs,mgc,osd_ldiskfs,fid,fld,lquota,ptlrpc
>
> Client modules:
> [adilger at twoshoes ~]$ lsmod | grep ^obdclass
> obdclass 1568040 86 osc,mgc,lustre,lov,mdc,fid,lmv,fld,ptlrpc
Thanks for this - a useful perspective.
I don't see quota (or lquota) on the client side though ??
>
>
> James, if we are going to move in the direction of having a separate
> "lustre-server" module, it would make sense to land a patch to master
> now that adds both a module alias, as well as a filesystem type alias
> for lustre-server, so that we can start moving systems over to using
> "lustre-server" as the filesystem type in /etc/fstab or HA mounting
> scripts, and it will auto-load the module at the same time. This might
> also allow the Lustre mount code to be simplified, since I think it has
> some messy code to have the same code, but distinguish between client
> and server mounts.
This sounds like an excellent idea!
I have a patch in my lustre-testing tree (probably will post next week)
which moves the client-side mounting from obdclass to llite. It is a
nice simplification.
Thanks,
NeilBrown
>
> Cheers, Andreas
> ---
> Andreas Dilger
> Principal Lustre Architect
> Whamcloud
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180628/5d3ab435/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-27 12:06 ` Patrick Farrell
@ 2018-06-28 1:59 ` NeilBrown
2018-06-28 2:35 ` Patrick Farrell
2018-06-28 15:03 ` Cory Spitz
0 siblings, 2 replies; 82+ messages in thread
From: NeilBrown @ 2018-06-28 1:59 UTC (permalink / raw)
To: lustre-devel
On Wed, Jun 27 2018, Patrick Farrell wrote:
> Neil,
>
> We do indeed have such functionality (it?s called DVS and it?s
> basically a high speed file system projection framework, ala NFS but
> faster), so the ability to build lnet separately is valuable to us.
> While it is being open sourced under the GPL, I don?t think there?s
> any intention to try to upstream it. The current code isn?t even
> usable off of Cray systems as it depends on info from user space (that
> is provided, in the end, from Cray proprietary hardware) to keep its
> connection/routing tables up to date. That?s supposedly in the
> pipeline to get fixed, but it?s still pretty far from generally
> usable.
>
> But we?d still really appreciate it if lnet stayed separate. Don?t
> know if that?s enough for you - I know sometimes *small* stuff is done
> for out of tree users. Hopefully this meets that standard.
>
Ahh - DVS. That answers a question I just asked in another email.
My google-skills don't seem to be up to locating the source code though
:-(
While I wouldn't knowingly break an interface used by some out-of-tree
code without good reason, it is hard to avoid if you don't know what the
out-of-tree code does. It can be very tempting to remove something that
isn't being used, but that can certainly hurt out-of-tree code
sometimes.
A particular example I'm exploring at present is the dual data paths in
LNet. Or maybe it is dual types of Memory Descriptors.
There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
uses page+offset.
The kiov option isn't used in the client code and it seems likely that
the server-side code could be converted to use iovec without problems.
I'd like to remove the kiov as I wouldn't be able to justify its
existence when submitting the client-only code upstream. But I don't
want to remove the option of having an alternate MD type if it really is
significantly more efficient in some context.
If I know whether DVS used kiov or iovec - and in what way - that would
help me to know if I might break something, and to be able to assess the
cost.
In my mind, the "standard" that you mention is always about
practicality. Code needs to be maintainable - easy to understand and
hard to break. If the LNet interface is clean and well documented in
the kernel, then I don't see why we would not at least attempt to
preserve it.
Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180628/d261f326/attachment-0001.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-28 1:59 ` NeilBrown
@ 2018-06-28 2:35 ` Patrick Farrell
2018-06-28 23:12 ` Andreas Dilger
2018-06-28 15:03 ` Cory Spitz
1 sibling, 1 reply; 82+ messages in thread
From: Patrick Farrell @ 2018-06-28 2:35 UTC (permalink / raw)
To: lustre-devel
I will ask about source and plans to publish it, Neil - I?m almost certain it has not been publicly released yet and I?m not actually sure the relicensing is really complete. So I?d be shocked and possibly mildly concerned if you found the source online today.
Understood about the rest, and thanks! I have no idea which DVS uses, and I understand that as an out of tree user, we are in the end responsible for it. DVS has in fact had this relationship with the master branch for a long time - Andreas and company are accommodating about interface changes, but in the end, we are responsible for keeping in sync.
I would say about the acceptor and sock lnd: I believe Lustre assumes some IP transport is available for configuration, but does NOT necessarily use it for primary communication. Fabrics - Like infiniband or Cray Aries - are more or less always configured to provide IP transport, to enable the panoply of tools and apps that rely on it. But they perform better if their native protocols are used, which is of course what the other LNDs do.
________________________________
From: NeilBrown <neilb@suse.com>
Sent: Wednesday, June 27, 2018 8:59:05 PM
To: Patrick Farrell; Andreas Dilger; James Simmons
Cc: Oleg Drokin; Lustre Development List
Subject: Re: [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
On Wed, Jun 27 2018, Patrick Farrell wrote:
> Neil,
>
> We do indeed have such functionality (it?s called DVS and it?s
> basically a high speed file system projection framework, ala NFS but
> faster), so the ability to build lnet separately is valuable to us.
> While it is being open sourced under the GPL, I don?t think there?s
> any intention to try to upstream it. The current code isn?t even
> usable off of Cray systems as it depends on info from user space (that
> is provided, in the end, from Cray proprietary hardware) to keep its
> connection/routing tables up to date. That?s supposedly in the
> pipeline to get fixed, but it?s still pretty far from generally
> usable.
>
> But we?d still really appreciate it if lnet stayed separate. Don?t
> know if that?s enough for you - I know sometimes *small* stuff is done
> for out of tree users. Hopefully this meets that standard.
>
Ahh - DVS. That answers a question I just asked in another email.
My google-skills don't seem to be up to locating the source code though
:-(
While I wouldn't knowingly break an interface used by some out-of-tree
code without good reason, it is hard to avoid if you don't know what the
out-of-tree code does. It can be very tempting to remove something that
isn't being used, but that can certainly hurt out-of-tree code
sometimes.
A particular example I'm exploring at present is the dual data paths in
LNet. Or maybe it is dual types of Memory Descriptors.
There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
uses page+offset.
The kiov option isn't used in the client code and it seems likely that
the server-side code could be converted to use iovec without problems.
I'd like to remove the kiov as I wouldn't be able to justify its
existence when submitting the client-only code upstream. But I don't
want to remove the option of having an alternate MD type if it really is
significantly more efficient in some context.
If I know whether DVS used kiov or iovec - and in what way - that would
help me to know if I might break something, and to be able to assess the
cost.
In my mind, the "standard" that you mention is always about
practicality. Code needs to be maintainable - easy to understand and
hard to break. If the LNet interface is clean and well documented in
the kernel, then I don't see why we would not at least attempt to
preserve it.
Thanks,
NeilBrown
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180628/6750beca/attachment.html>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-28 1:59 ` NeilBrown
2018-06-28 2:35 ` Patrick Farrell
@ 2018-06-28 15:03 ` Cory Spitz
2018-06-28 17:03 ` Doug Oucharek
1 sibling, 1 reply; 82+ messages in thread
From: Cory Spitz @ 2018-06-28 15:03 UTC (permalink / raw)
To: lustre-devel
I know that Doug follows this list closely, but I'm going to flag him here directly, because I know that he has an opinion about kiov / iovec (and a 3rd if I recall correctly).
Doug told me that (possibly) Al Viro had a proposal or submitted code to simply the kiov usage and that Doug was in favor of it. Let's choose to make the right architectural call for Lustre/LNet and let Cray worry about keeping our out-of-tree code working. As Patrick said, Cray pays the price of keeping DVS off to the side (we're wising up, which is why it is now GPL... but don't expect us to publish it on the Internet necessarily... baby-steps).
Doug, can you weigh-in on this case?
Thanks,
-Cory
--
?On 6/27/18, 8:59 PM, "lustre-devel on behalf of NeilBrown" <lustre-devel-bounces at lists.lustre.org on behalf of neilb@suse.com> wrote:
On Wed, Jun 27 2018, Patrick Farrell wrote:
> Neil,
>
> We do indeed have such functionality (it?s called DVS and it?s
> basically a high speed file system projection framework, ala NFS but
> faster), so the ability to build lnet separately is valuable to us.
> While it is being open sourced under the GPL, I don?t think there?s
> any intention to try to upstream it. The current code isn?t even
> usable off of Cray systems as it depends on info from user space (that
> is provided, in the end, from Cray proprietary hardware) to keep its
> connection/routing tables up to date. That?s supposedly in the
> pipeline to get fixed, but it?s still pretty far from generally
> usable.
>
> But we?d still really appreciate it if lnet stayed separate. Don?t
> know if that?s enough for you - I know sometimes *small* stuff is done
> for out of tree users. Hopefully this meets that standard.
>
Ahh - DVS. That answers a question I just asked in another email.
My google-skills don't seem to be up to locating the source code though
:-(
While I wouldn't knowingly break an interface used by some out-of-tree
code without good reason, it is hard to avoid if you don't know what the
out-of-tree code does. It can be very tempting to remove something that
isn't being used, but that can certainly hurt out-of-tree code
sometimes.
A particular example I'm exploring at present is the dual data paths in
LNet. Or maybe it is dual types of Memory Descriptors.
There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
uses page+offset.
The kiov option isn't used in the client code and it seems likely that
the server-side code could be converted to use iovec without problems.
I'd like to remove the kiov as I wouldn't be able to justify its
existence when submitting the client-only code upstream. But I don't
want to remove the option of having an alternate MD type if it really is
significantly more efficient in some context.
If I know whether DVS used kiov or iovec - and in what way - that would
help me to know if I might break something, and to be able to assess the
cost.
In my mind, the "standard" that you mention is always about
practicality. Code needs to be maintainable - easy to understand and
hard to break. If the LNet interface is clean and well documented in
the kernel, then I don't see why we would not at least attempt to
preserve it.
Thanks,
NeilBrown
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-28 15:03 ` Cory Spitz
@ 2018-06-28 17:03 ` Doug Oucharek
2018-07-04 17:54 ` Alexey Lyashkov
0 siblings, 1 reply; 82+ messages in thread
From: Doug Oucharek @ 2018-06-28 17:03 UTC (permalink / raw)
To: lustre-devel
Yes, Al Viro had started the process of getting Lustre in the upstream staging area to use iov_iter and, I believe, biovec. A very good change in my opinion as it gets rid of the duality. Unfortunately, those changes have not made their way into the Whamcloud tree yet so there is a disconnect when working between the two trees. James Simmons might know more, but I don?t believe Al?s changes are complete so there is still a need to support iovec and kvec. Could we replace these two with biovecs? James?
As Cory says, lets make the right architectural decisions and let companies (Whamcloud, Cray, etc) absorb the improvements. We can adapt DVS as required.
As for LNet as a module: please keep it as a separate module. Merging it into Lustre would create a dependency for DVS to have Lustre installed and possibly running. This would also stand in the way of any potential future users of LNet. Andreas pointed out one, Zest (search for LNET in this paper: https://www.psc.edu/images/zest/zest-pdsw08-paper.pdf), but there may be others developing. LNet could become its own project at some point because of multiple users.
Doug
> On Jun 28, 2018, at 8:03 AM, Cory Spitz <spitzcor@cray.com> wrote:
>
> I know that Doug follows this list closely, but I'm going to flag him here directly, because I know that he has an opinion about kiov / iovec (and a 3rd if I recall correctly).
>
> Doug told me that (possibly) Al Viro had a proposal or submitted code to simply the kiov usage and that Doug was in favor of it. Let's choose to make the right architectural call for Lustre/LNet and let Cray worry about keeping our out-of-tree code working. As Patrick said, Cray pays the price of keeping DVS off to the side (we're wising up, which is why it is now GPL... but don't expect us to publish it on the Internet necessarily... baby-steps).
>
> Doug, can you weigh-in on this case?
>
> Thanks,
> -Cory
>
> --
>
> ?On 6/27/18, 8:59 PM, "lustre-devel on behalf of NeilBrown" <lustre-devel-bounces at lists.lustre.org on behalf of neilb@suse.com> wrote:
>
> On Wed, Jun 27 2018, Patrick Farrell wrote:
>
>> Neil,
>>
>> We do indeed have such functionality (it?s called DVS and it?s
>> basically a high speed file system projection framework, ala NFS but
>> faster), so the ability to build lnet separately is valuable to us.
>> While it is being open sourced under the GPL, I don?t think there?s
>> any intention to try to upstream it. The current code isn?t even
>> usable off of Cray systems as it depends on info from user space (that
>> is provided, in the end, from Cray proprietary hardware) to keep its
>> connection/routing tables up to date. That?s supposedly in the
>> pipeline to get fixed, but it?s still pretty far from generally
>> usable.
>>
>> But we?d still really appreciate it if lnet stayed separate. Don?t
>> know if that?s enough for you - I know sometimes *small* stuff is done
>> for out of tree users. Hopefully this meets that standard.
>>
>
> Ahh - DVS. That answers a question I just asked in another email.
> My google-skills don't seem to be up to locating the source code though
> :-(
>
> While I wouldn't knowingly break an interface used by some out-of-tree
> code without good reason, it is hard to avoid if you don't know what the
> out-of-tree code does. It can be very tempting to remove something that
> isn't being used, but that can certainly hurt out-of-tree code
> sometimes.
>
> A particular example I'm exploring at present is the dual data paths in
> LNet. Or maybe it is dual types of Memory Descriptors.
> There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
> uses page+offset.
> The kiov option isn't used in the client code and it seems likely that
> the server-side code could be converted to use iovec without problems.
>
> I'd like to remove the kiov as I wouldn't be able to justify its
> existence when submitting the client-only code upstream. But I don't
> want to remove the option of having an alternate MD type if it really is
> significantly more efficient in some context.
> If I know whether DVS used kiov or iovec - and in what way - that would
> help me to know if I might break something, and to be able to assess the
> cost.
>
> In my mind, the "standard" that you mention is always about
> practicality. Code needs to be maintainable - easy to understand and
> hard to break. If the LNet interface is clean and well documented in
> the kernel, then I don't see why we would not at least attempt to
> preserve it.
>
> Thanks,
> NeilBrown
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-28 2:35 ` Patrick Farrell
@ 2018-06-28 23:12 ` Andreas Dilger
2018-07-05 23:47 ` James Simmons
0 siblings, 1 reply; 82+ messages in thread
From: Andreas Dilger @ 2018-06-28 23:12 UTC (permalink / raw)
To: lustre-devel
On Jun 27, 2018, at 20:35, Patrick Farrell <paf@cray.com> wrote:
>
> I would say about the acceptor and sock lnd: I believe Lustre assumes some IP transport is available for configuration, but does NOT necessarily use it for primary communication. Fabrics - Like infiniband or Cray Aries - are more or less always configured to provide IP transport, to enable the panoply of tools and apps that rely on it. But they perform better if their native protocols are used, which is of course what the other LNDs do.
It is worthwhile to clarify this a bit - LNet uses IP for *addressing* of the nodes at connection time, but I don't think it even uses the TCP interface for any communication itself (though I could be mistaken, as LNet isn't my specialty). After the initial connection, the only other thing the IP addresses are used for is printing in error messages.
At one time or another we've discussed how we might get rid of the need for IPoIB on client nodes, since some sites don't want any IP connectivity to the client nodes for security and performance reasons. That said, I don't think we've come up with a good solution yet. LNet itself allows alternate addressing schemes to be used. The former qswlnd (for Quadrics Elan networks) just used an integer node number, like 1 at elan for the NID, but I don't think there is any such alternative for IB node addresses except using a MAC hardware address or similar.
Cheers, Andreas
>
> Neil Brown <neilb@suse.com> wrote:
>>
>> On Wed, Jun 27 2018, Patrick Farrell wrote:
>>
>> > Neil,
>> >
>> > We do indeed have such functionality (it?s called DVS and it?s
>> > basically a high speed file system projection framework, ala NFS but
>> > faster), so the ability to build lnet separately is valuable to us.
>> > While it is being open sourced under the GPL, I don?t think there?s
>> > any intention to try to upstream it. The current code isn?t even
>> > usable off of Cray systems as it depends on info from user space (that
>> > is provided, in the end, from Cray proprietary hardware) to keep its
>> > connection/routing tables up to date. That?s supposedly in the
>> > pipeline to get fixed, but it?s still pretty far from generally
>> > usable.
>> >
>> > But we?d still really appreciate it if lnet stayed separate. Don?t
>> > know if that?s enough for you - I know sometimes *small* stuff is done
>> > for out of tree users. Hopefully this meets that standard.
>> >
>>
>> Ahh - DVS. That answers a question I just asked in another email.
>> My google-skills don't seem to be up to locating the source code though
>> :-(
>>
>> While I wouldn't knowingly break an interface used by some out-of-tree
>> code without good reason, it is hard to avoid if you don't know what the
>> out-of-tree code does. It can be very tempting to remove something that
>> isn't being used, but that can certainly hurt out-of-tree code
>> sometimes.
>>
>> A particular example I'm exploring at present is the dual data paths in
>> LNet. Or maybe it is dual types of Memory Descriptors.
>> There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
>> uses page+offset.
>> The kiov option isn't used in the client code and it seems likely that
>> the server-side code could be converted to use iovec without problems.
>>
>> I'd like to remove the kiov as I wouldn't be able to justify its
>> existence when submitting the client-only code upstream. But I don't
>> want to remove the option of having an alternate MD type if it really is
>> significantly more efficient in some context.
>> If I know whether DVS used kiov or iovec - and in what way - that would
>> help me to know if I might break something, and to be able to assess the
>> cost.
>>
>> In my mind, the "standard" that you mention is always about
>> practicality. Code needs to be maintainable - easy to understand and
>> hard to break. If the LNet interface is clean and well documented in
>> the kernel, then I don't see why we would not at least attempt to
>> preserve it.
>>
>> Thanks,
>> NeilBrown
Cheers, Andreas
---
Andreas Dilger
Principal Lustre Architect
Whamcloud
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 235 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180628/9e2591e1/attachment.sig>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-28 17:03 ` Doug Oucharek
@ 2018-07-04 17:54 ` Alexey Lyashkov
0 siblings, 0 replies; 82+ messages in thread
From: Alexey Lyashkov @ 2018-07-04 17:54 UTC (permalink / raw)
To: lustre-devel
Doug,
What about LNet API change to use sge list? It's more natively structure for variable page list, which is may a short sized for MDC links and full sized for OST links.
?On 28/06/2018, 20:04, "Doug Oucharek" <doucharek@cray.com> wrote:
Yes, Al Viro had started the process of getting Lustre in the upstream staging area to use iov_iter and, I believe, biovec. A very good change in my opinion as it gets rid of the duality. Unfortunately, those changes have not made their way into the Whamcloud tree yet so there is a disconnect when working between the two trees. James Simmons might know more, but I don?t believe Al?s changes are complete so there is still a need to support iovec and kvec. Could we replace these two with biovecs? James?
As Cory says, lets make the right architectural decisions and let companies (Whamcloud, Cray, etc) absorb the improvements. We can adapt DVS as required.
As for LNet as a module: please keep it as a separate module. Merging it into Lustre would create a dependency for DVS to have Lustre installed and possibly running. This would also stand in the way of any potential future users of LNet. Andreas pointed out one, Zest (search for LNET in this paper: https://www.psc.edu/images/zest/zest-pdsw08-paper.pdf), but there may be others developing. LNet could become its own project at some point because of multiple users.
Doug
> On Jun 28, 2018, at 8:03 AM, Cory Spitz <spitzcor@cray.com> wrote:
>
> I know that Doug follows this list closely, but I'm going to flag him here directly, because I know that he has an opinion about kiov / iovec (and a 3rd if I recall correctly).
>
> Doug told me that (possibly) Al Viro had a proposal or submitted code to simply the kiov usage and that Doug was in favor of it. Let's choose to make the right architectural call for Lustre/LNet and let Cray worry about keeping our out-of-tree code working. As Patrick said, Cray pays the price of keeping DVS off to the side (we're wising up, which is why it is now GPL... but don't expect us to publish it on the Internet necessarily... baby-steps).
>
> Doug, can you weigh-in on this case?
>
> Thanks,
> -Cory
>
> --
>
> On 6/27/18, 8:59 PM, "lustre-devel on behalf of NeilBrown" <lustre-devel-bounces at lists.lustre.org on behalf of neilb@suse.com> wrote:
>
> On Wed, Jun 27 2018, Patrick Farrell wrote:
>
>> Neil,
>>
>> We do indeed have such functionality (it?s called DVS and it?s
>> basically a high speed file system projection framework, ala NFS but
>> faster), so the ability to build lnet separately is valuable to us.
>> While it is being open sourced under the GPL, I don?t think there?s
>> any intention to try to upstream it. The current code isn?t even
>> usable off of Cray systems as it depends on info from user space (that
>> is provided, in the end, from Cray proprietary hardware) to keep its
>> connection/routing tables up to date. That?s supposedly in the
>> pipeline to get fixed, but it?s still pretty far from generally
>> usable.
>>
>> But we?d still really appreciate it if lnet stayed separate. Don?t
>> know if that?s enough for you - I know sometimes *small* stuff is done
>> for out of tree users. Hopefully this meets that standard.
>>
>
> Ahh - DVS. That answers a question I just asked in another email.
> My google-skills don't seem to be up to locating the source code though
> :-(
>
> While I wouldn't knowingly break an interface used by some out-of-tree
> code without good reason, it is hard to avoid if you don't know what the
> out-of-tree code does. It can be very tempting to remove something that
> isn't being used, but that can certainly hurt out-of-tree code
> sometimes.
>
> A particular example I'm exploring at present is the dual data paths in
> LNet. Or maybe it is dual types of Memory Descriptors.
> There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
> uses page+offset.
> The kiov option isn't used in the client code and it seems likely that
> the server-side code could be converted to use iovec without problems.
>
> I'd like to remove the kiov as I wouldn't be able to justify its
> existence when submitting the client-only code upstream. But I don't
> want to remove the option of having an alternate MD type if it really is
> significantly more efficient in some context.
> If I know whether DVS used kiov or iovec - and in what way - that would
> help me to know if I might break something, and to be able to assess the
> cost.
>
> In my mind, the "standard" that you mention is always about
> practicality. Code needs to be maintainable - easy to understand and
> hard to break. If the LNet interface is clean and well documented in
> the kernel, then I don't see why we would not at least attempt to
> preserve it.
>
> Thanks,
> NeilBrown
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-06-28 23:12 ` Andreas Dilger
@ 2018-07-05 23:47 ` James Simmons
2018-07-06 0:01 ` Doug Oucharek
0 siblings, 1 reply; 82+ messages in thread
From: James Simmons @ 2018-07-05 23:47 UTC (permalink / raw)
To: lustre-devel
> > I would say about the acceptor and sock lnd: I believe Lustre assumes some IP transport is available for configuration, but does NOT necessarily use it for primary communication. Fabrics - Like infiniband or Cray Aries - are more or less always configured to provide IP transport, to enable the panoply of tools and apps that rely on it. But they perform better if their native protocols are used, which is of course what the other LNDs do.
>
> It is worthwhile to clarify this a bit - LNet uses IP for *addressing* of the nodes at connection time, but I don't think it even uses the TCP interface for any communication itself (though I could be mistaken, as LNet isn't my specialty). After the initial connection, the only other thing the IP addresses are used for is printing in error messages.
>
> At one time or another we've discussed how we might get rid of the need for IPoIB on client nodes, since some sites don't want any IP connectivity to the client nodes for security and performance reasons. That said, I don't think we've come up with a good solution yet. LNet itself allows alternate addressing schemes to be used. The former qswlnd (for Quadrics Elan networks) just used an integer node number, like 1 at elan for the NID, but I don't think there is any such alternative for IB node addresses except using a MAC hardware address or similar.
The challenge to moving away from IP address for addressing is the NID
address itself. Currently it is 64 bits which is composed for two parts.
static inline lnet_nid_t LNET_MKNID(__u32 net, __u32 addr)
{
return (((__u64)net) << 32) | addr;
}
To use IB harware addressing would require a full 64 bit space so
lnet_nid_t is to small. This is reason IPv6 is not supported. Ideas
have floated around on how to get around this but nothing yet.
It would be a massive API change currently.
> > Neil Brown <neilb@suse.com> wrote:
> >>
> >> On Wed, Jun 27 2018, Patrick Farrell wrote:
> >>
> >> > Neil,
> >> >
> >> > We do indeed have such functionality (it?s called DVS and it?s
> >> > basically a high speed file system projection framework, ala NFS but
> >> > faster), so the ability to build lnet separately is valuable to us.
> >> > While it is being open sourced under the GPL, I don?t think there?s
> >> > any intention to try to upstream it. The current code isn?t even
> >> > usable off of Cray systems as it depends on info from user space (that
> >> > is provided, in the end, from Cray proprietary hardware) to keep its
> >> > connection/routing tables up to date. That?s supposedly in the
> >> > pipeline to get fixed, but it?s still pretty far from generally
> >> > usable.
> >> >
> >> > But we?d still really appreciate it if lnet stayed separate. Don?t
> >> > know if that?s enough for you - I know sometimes *small* stuff is done
> >> > for out of tree users. Hopefully this meets that standard.
> >> >
> >>
> >> Ahh - DVS. That answers a question I just asked in another email.
> >> My google-skills don't seem to be up to locating the source code though
> >> :-(
> >>
> >> While I wouldn't knowingly break an interface used by some out-of-tree
> >> code without good reason, it is hard to avoid if you don't know what the
> >> out-of-tree code does. It can be very tempting to remove something that
> >> isn't being used, but that can certainly hurt out-of-tree code
> >> sometimes.
> >>
> >> A particular example I'm exploring at present is the dual data paths in
> >> LNet. Or maybe it is dual types of Memory Descriptors.
> >> There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
> >> uses page+offset.
> >> The kiov option isn't used in the client code and it seems likely that
> >> the server-side code could be converted to use iovec without problems.
> >>
> >> I'd like to remove the kiov as I wouldn't be able to justify its
> >> existence when submitting the client-only code upstream. But I don't
> >> want to remove the option of having an alternate MD type if it really is
> >> significantly more efficient in some context.
> >> If I know whether DVS used kiov or iovec - and in what way - that would
> >> help me to know if I might break something, and to be able to assess the
> >> cost.
> >>
> >> In my mind, the "standard" that you mention is always about
> >> practicality. Code needs to be maintainable - easy to understand and
> >> hard to break. If the LNet interface is clean and well documented in
> >> the kernel, then I don't see why we would not at least attempt to
> >> preserve it.
> >>
> >> Thanks,
> >> NeilBrown
>
> Cheers, Andreas
> ---
> Andreas Dilger
> Principal Lustre Architect
> Whamcloud
>
>
>
>
>
>
>
>
^ permalink raw reply [flat|nested] 82+ messages in thread
* [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction.
2018-07-05 23:47 ` James Simmons
@ 2018-07-06 0:01 ` Doug Oucharek
0 siblings, 0 replies; 82+ messages in thread
From: Doug Oucharek @ 2018-07-06 0:01 UTC (permalink / raw)
To: lustre-devel
Unfortunately, NIDs get stored outside of LNet in the Lustre configuration. That?s why changing the nature of NIDs requires a lot of changes in many places in and out of LNet.
In my opinion, the best gameplay is to get Lustre to stop even knowing what a NID is. Have Lustre identify nodes via a name which LNet converts to a NID (sort of like how DNS works). At that point, LNet can make the NID be anything you want: IPv6, IB address, etc. This also means you don?t have to run writeconf (or something like that) when the network changes. Only LNet needs to be updated.
Doug
> On Jul 5, 2018, at 4:47 PM, James Simmons <jsimmons@infradead.org> wrote:
>
>
>>> I would say about the acceptor and sock lnd: I believe Lustre assumes some IP transport is available for configuration, but does NOT necessarily use it for primary communication. Fabrics - Like infiniband or Cray Aries - are more or less always configured to provide IP transport, to enable the panoply of tools and apps that rely on it. But they perform better if their native protocols are used, which is of course what the other LNDs do.
>>
>> It is worthwhile to clarify this a bit - LNet uses IP for *addressing* of the nodes at connection time, but I don't think it even uses the TCP interface for any communication itself (though I could be mistaken, as LNet isn't my specialty). After the initial connection, the only other thing the IP addresses are used for is printing in error messages.
>>
>> At one time or another we've discussed how we might get rid of the need for IPoIB on client nodes, since some sites don't want any IP connectivity to the client nodes for security and performance reasons. That said, I don't think we've come up with a good solution yet. LNet itself allows alternate addressing schemes to be used. The former qswlnd (for Quadrics Elan networks) just used an integer node number, like 1 at elan for the NID, but I don't think there is any such alternative for IB node addresses except using a MAC hardware address or similar.
>
> The challenge to moving away from IP address for addressing is the NID
> address itself. Currently it is 64 bits which is composed for two parts.
>
> static inline lnet_nid_t LNET_MKNID(__u32 net, __u32 addr)
> {
> return (((__u64)net) << 32) | addr;
> }
>
> To use IB harware addressing would require a full 64 bit space so
> lnet_nid_t is to small. This is reason IPv6 is not supported. Ideas
> have floated around on how to get around this but nothing yet.
> It would be a massive API change currently.
>
>>> Neil Brown <neilb@suse.com> wrote:
>>>>
>>>> On Wed, Jun 27 2018, Patrick Farrell wrote:
>>>>
>>>>> Neil,
>>>>>
>>>>> We do indeed have such functionality (it?s called DVS and it?s
>>>>> basically a high speed file system projection framework, ala NFS but
>>>>> faster), so the ability to build lnet separately is valuable to us.
>>>>> While it is being open sourced under the GPL, I don?t think there?s
>>>>> any intention to try to upstream it. The current code isn?t even
>>>>> usable off of Cray systems as it depends on info from user space (that
>>>>> is provided, in the end, from Cray proprietary hardware) to keep its
>>>>> connection/routing tables up to date. That?s supposedly in the
>>>>> pipeline to get fixed, but it?s still pretty far from generally
>>>>> usable.
>>>>>
>>>>> But we?d still really appreciate it if lnet stayed separate. Don?t
>>>>> know if that?s enough for you - I know sometimes *small* stuff is done
>>>>> for out of tree users. Hopefully this meets that standard.
>>>>>
>>>>
>>>> Ahh - DVS. That answers a question I just asked in another email.
>>>> My google-skills don't seem to be up to locating the source code though
>>>> :-(
>>>>
>>>> While I wouldn't knowingly break an interface used by some out-of-tree
>>>> code without good reason, it is hard to avoid if you don't know what the
>>>> out-of-tree code does. It can be very tempting to remove something that
>>>> isn't being used, but that can certainly hurt out-of-tree code
>>>> sometimes.
>>>>
>>>> A particular example I'm exploring at present is the dual data paths in
>>>> LNet. Or maybe it is dual types of Memory Descriptors.
>>>> There is 'kiov' which uses kernel-virtual addresses and 'iovec' which
>>>> uses page+offset.
>>>> The kiov option isn't used in the client code and it seems likely that
>>>> the server-side code could be converted to use iovec without problems.
>>>>
>>>> I'd like to remove the kiov as I wouldn't be able to justify its
>>>> existence when submitting the client-only code upstream. But I don't
>>>> want to remove the option of having an alternate MD type if it really is
>>>> significantly more efficient in some context.
>>>> If I know whether DVS used kiov or iovec - and in what way - that would
>>>> help me to know if I might break something, and to be able to assess the
>>>> cost.
>>>>
>>>> In my mind, the "standard" that you mention is always about
>>>> practicality. Code needs to be maintainable - easy to understand and
>>>> hard to break. If the LNet interface is clean and well documented in
>>>> the kernel, then I don't see why we would not at least attempt to
>>>> preserve it.
>>>>
>>>> Thanks,
>>>> NeilBrown
>>
>> Cheers, Andreas
>> ---
>> Andreas Dilger
>> Principal Lustre Architect
>> Whamcloud
>>
>>
>>
>>
>>
>>
>>
> _______________________________________________
> lustre-devel mailing list
> lustre-devel at lists.lustre.org
> http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org
^ permalink raw reply [flat|nested] 82+ messages in thread
end of thread, other threads:[~2018-07-06 0:01 UTC | newest]
Thread overview: 82+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-15 7:11 [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 18/24] lustre/lnet: move tracefile locking from linux-tracefile.c to tracefile.c NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 17/24] lustre: remove lustre_compat.h NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 21/24] lustre: move remainder of linux-tracefile.c to tracefile.c NeilBrown
2018-06-21 1:37 ` James Simmons
2018-06-22 3:19 ` NeilBrown
2018-06-24 20:29 ` James Simmons
2018-06-24 23:25 ` NeilBrown
2018-06-25 21:52 ` James Simmons
2018-06-15 7:11 ` [lustre-devel] [PATCH 03/24] kbuild: support building of per-directory mod.a NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 16/24] lustre: remove redefinition of module_init() NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 06/24] lustre: build ldlm in the ldlm directory NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 05/24] kbuild: Add documentation for modobj-m NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 11/24] lustre: discard current_n*groups macros NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 04/24] kbuild: disable KBUILD_MODNAME when building for mod.a NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 24/24] lustre: discard TCD_MAX_TYPES NeilBrown
2018-06-21 1:38 ` James Simmons
2018-06-21 4:28 ` NeilBrown
2018-06-24 20:37 ` James Simmons
2018-06-24 23:26 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 13/24] lustre: discard NO_QUOTA and QUOTA_OK NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 01/24] kbuild: detect directories in components of a module NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 22/24] lustre: fold cfs_tracefile_*_arch into their only callers NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 08/24] lustre: use truncate_inode_page in place of truncate_complete_page NeilBrown
2018-06-16 0:17 ` James Simmons
2018-06-16 0:52 ` NeilBrown
2018-06-16 0:59 ` Oleg Drokin
2018-06-18 2:03 ` NeilBrown
2018-06-25 0:02 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() NeilBrown
2018-06-25 0:02 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages NeilBrown
2018-06-25 0:03 ` [lustre-devel] [PATCH 3/3] lustre: discard truncate_complete_page() NeilBrown
2018-06-26 0:19 ` James Simmons
2018-06-25 0:38 ` [lustre-devel] [PATCH 2/3] lustre: use delete_from_page_cache() for directory pages Oleg Drokin
2018-06-26 0:18 ` James Simmons
2018-06-25 0:38 ` [lustre-devel] [PATCH 1/3] lustre: use generic_error_remove_page() Oleg Drokin
2018-06-25 1:57 ` NeilBrown
2018-06-26 0:26 ` James Simmons
2018-06-26 0:14 ` James Simmons
2018-06-15 7:11 ` [lustre-devel] [PATCH 15/24] lustre: centralize TIMES_SET_FLAGS NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 07/24] lustre: merge libcfs module into lnet NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 09/24] lustre: don't use spare bits in iattr.ia_valid NeilBrown
2018-06-21 1:35 ` James Simmons
2018-06-21 4:19 ` NeilBrown
2018-06-22 2:23 ` NeilBrown
2018-06-24 20:33 ` James Simmons
2018-06-24 23:50 ` NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 20/24] lustre: move tcd locking across to tracefile.c NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 10/24] lustre: remove lustre_patchless_compat.h NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 19/24] lustre: open code cfs_trace_max_debug_mb() into cfs_trace_set_debug_mb() NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 02/24] kbuild: treat a directory listed in a composite object as foo/mod.a NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 14/24] lustre: discard ext2* bit operations NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 12/24] lustre: discard LTIME_S macro NeilBrown
2018-06-15 7:11 ` [lustre-devel] [PATCH 23/24] lustre: renamed CFS_TCD_TYPE_MAX to CFS_TCD_TYPE_CNT NeilBrown
2018-06-21 1:48 ` [lustre-devel] [PATCH 00/24] lustre - more cleanups including module reduction James Simmons
2018-06-21 2:29 ` Patrick Farrell
2018-06-21 2:57 ` James Simmons
2018-06-21 7:22 ` Andreas Dilger
2018-06-21 12:28 ` Patrick Farrell
2018-06-24 20:35 ` James Simmons
2018-06-24 23:40 ` NeilBrown
2018-06-26 1:13 ` James Simmons
2018-06-26 13:51 ` Patrick Farrell
2018-06-27 3:08 ` NeilBrown
2018-06-27 4:00 ` Cory Spitz
2018-06-28 1:26 ` NeilBrown
2018-06-27 11:01 ` Andreas Dilger
2018-06-27 12:06 ` Patrick Farrell
2018-06-28 1:59 ` NeilBrown
2018-06-28 2:35 ` Patrick Farrell
2018-06-28 23:12 ` Andreas Dilger
2018-07-05 23:47 ` James Simmons
2018-07-06 0:01 ` Doug Oucharek
2018-06-28 15:03 ` Cory Spitz
2018-06-28 17:03 ` Doug Oucharek
2018-07-04 17:54 ` Alexey Lyashkov
2018-06-28 1:39 ` NeilBrown
2018-06-21 3:00 ` NeilBrown
2018-06-21 3:42 ` Patrick Farrell
2018-06-24 20:39 ` James Simmons
2018-06-21 2:54 ` NeilBrown
2018-06-24 20:46 ` James Simmons
2018-06-24 23:58 ` NeilBrown
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.