* [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches @ 2021-09-02 21:57 Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 1/6] multipath.conf: fix typo in ghost_delay description Benjamin Marzinski ` (5 more replies) 0 siblings, 6 replies; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck Here are a couple of unconnected patches. They (at least libmultipath.version) are meant to be applied on top of Martin's recent patchset. Changes from v1: 3/6: Reworked this to keep the logic in multipath, and prefix the output with "Warning" as suggested by Martin Wilck. 6/6: New patch to simplify code that I noticed while reworking patch 3/6. Benjamin Marzinski (6): multipath.conf: fix typo in ghost_delay description mpathpersist: fail commands when no usable paths exist multipath: print warning if multipathd is not running. libmultipath: remove unneeded code in coalesce_paths libmultipath: deal with dynamic PTHREAD_STACK_MIN libmultipath: drop unnecessary parameter from remove_map() libmpathpersist/mpath_persist.c | 10 ++++- libmultipath/configure.c | 62 ++++--------------------------- libmultipath/configure.h | 1 + libmultipath/libmultipath.version | 5 +++ libmultipath/structs_vec.c | 12 +++--- libmultipath/structs_vec.h | 11 +----- libmultipath/util.c | 4 +- multipath/main.c | 18 +++++++-- multipath/multipath.conf.5 | 2 +- multipathd/dmevents.c | 2 +- multipathd/main.c | 12 +++--- 11 files changed, 54 insertions(+), 85 deletions(-) -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply [flat|nested] 11+ messages in thread
* [dm-devel] [PATCH v2 1/6] multipath.conf: fix typo in ghost_delay description 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski @ 2021-09-02 21:57 ` Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 2/6] mpathpersist: fail commands when no usable paths exist Benjamin Marzinski ` (4 subsequent siblings) 5 siblings, 0 replies; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck Reviewed-by: Martin Wilck <mwilck@suse.com> Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- multipath/multipath.conf.5 | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/multipath/multipath.conf.5 b/multipath/multipath.conf.5 index 064e4826..d6b8c7f6 100644 --- a/multipath/multipath.conf.5 +++ b/multipath/multipath.conf.5 @@ -1251,7 +1251,7 @@ The default is: in \fB/sys/block/<dev>/queue/max_sectors_kb\fR Sets the number of seconds that multipath will wait after creating a device with only ghost paths before marking it ready for use in systemd. This gives the active paths time to appear before the multipath runs the hardware handler -to switch the ghost paths to active ones. Setting this to \fI0\fR or \fIon\fR +to switch the ghost paths to active ones. Setting this to \fI0\fR or \fIno\fR makes multipath immediately mark a device with only ghost paths as ready. .RS .TP -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply related [flat|nested] 11+ messages in thread
* [dm-devel] [PATCH v2 2/6] mpathpersist: fail commands when no usable paths exist 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 1/6] multipath.conf: fix typo in ghost_delay description Benjamin Marzinski @ 2021-09-02 21:57 ` Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 3/6] multipath: print warning if multipathd is not running Benjamin Marzinski ` (3 subsequent siblings) 5 siblings, 0 replies; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck "mpathpersist -oCK <reservation_key> <device>" will return success if it is run on devices with no usable paths, but nothing is actually done. The -L command will fail, but it should give up sooner, and with a more helpful error message. Reviewed-by: Martin Wilck <mwilck@suse.com> Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- libmpathpersist/mpath_persist.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/libmpathpersist/mpath_persist.c b/libmpathpersist/mpath_persist.c index 190e9707..26710e79 100644 --- a/libmpathpersist/mpath_persist.c +++ b/libmpathpersist/mpath_persist.c @@ -604,7 +604,8 @@ int mpath_prout_common(struct multipath *mpp,int rq_servact, int rq_scope, return ret ; } } - return MPATH_PR_SUCCESS; + condlog (0, "%s: no path available", mpp->wwid); + return MPATH_PR_DMMP_ERROR; } int send_prout_activepath(char * dev, int rq_servact, int rq_scope, @@ -663,6 +664,11 @@ int mpath_prout_rel(struct multipath *mpp,int rq_servact, int rq_scope, active_pathcount = count_active_paths(mpp); + if (active_pathcount == 0) { + condlog (0, "%s: no path available", mpp->wwid); + return MPATH_PR_DMMP_ERROR; + } + struct threadinfo thread[active_pathcount]; memset(thread, 0, sizeof(thread)); for (i = 0; i < active_pathcount; i++){ -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply related [flat|nested] 11+ messages in thread
* [dm-devel] [PATCH v2 3/6] multipath: print warning if multipathd is not running. 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 1/6] multipath.conf: fix typo in ghost_delay description Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 2/6] mpathpersist: fail commands when no usable paths exist Benjamin Marzinski @ 2021-09-02 21:57 ` Benjamin Marzinski 2021-09-03 8:20 ` Martin Wilck 2021-09-02 21:57 ` [dm-devel] [PATCH v2 4/6] libmultipath: remove unneeded code in coalesce_paths Benjamin Marzinski ` (2 subsequent siblings) 5 siblings, 1 reply; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck If multipath notices that multipath devices exist or were created, and multipathd is not running, it now prints a warning message, so users are notified of the issue. Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- libmultipath/configure.c | 3 ++- libmultipath/configure.h | 1 + libmultipath/libmultipath.version | 5 +++++ multipath/main.c | 14 ++++++++++++-- 4 files changed, 20 insertions(+), 3 deletions(-) diff --git a/libmultipath/configure.c b/libmultipath/configure.c index 12278640..2f417914 100644 --- a/libmultipath/configure.c +++ b/libmultipath/configure.c @@ -1082,7 +1082,8 @@ deadmap (struct multipath * mpp) return 1; /* dead */ } -int check_daemon(void) +extern int +check_daemon(void) { int fd; char *reply; diff --git a/libmultipath/configure.h b/libmultipath/configure.h index 92a5aba6..efe18b7d 100644 --- a/libmultipath/configure.h +++ b/libmultipath/configure.h @@ -59,3 +59,4 @@ struct udev_device *get_udev_device(const char *dev, enum devtypes dev_type); void trigger_paths_udev_change(struct multipath *mpp, bool is_mpath); void trigger_partitions_udev_change(struct udev_device *dev, const char *action, int len); +int check_daemon(void); diff --git a/libmultipath/libmultipath.version b/libmultipath/libmultipath.version index 1d84d97c..7b48265f 100644 --- a/libmultipath/libmultipath.version +++ b/libmultipath/libmultipath.version @@ -285,3 +285,8 @@ global: print_strbuf; truncate_strbuf; } LIBMULTIPATH_8.0.0; + +LIBMULTIPATH_8.2.0 { +global: + check_daemon; +} LIBMULTIPATH_8.1.0; diff --git a/multipath/main.c b/multipath/main.c index 8fc0e15f..6c70760f 100644 --- a/multipath/main.c +++ b/multipath/main.c @@ -456,6 +456,7 @@ configure (struct config *conf, enum mpath_cmds cmd, { vector curmp = NULL; vector pathvec = NULL; + vector newmp = NULL; int r = RTVL_FAIL, rc; int di_flag = 0; char * refwwid = NULL; @@ -466,8 +467,9 @@ configure (struct config *conf, enum mpath_cmds cmd, */ curmp = vector_alloc(); pathvec = vector_alloc(); + newmp = vector_alloc(); - if (!curmp || !pathvec) { + if (!curmp || !pathvec || !newmp) { condlog(0, "can not allocate memory"); goto out; } @@ -569,16 +571,24 @@ configure (struct config *conf, enum mpath_cmds cmd, /* * core logic entry point */ - rc = coalesce_paths(&vecs, NULL, refwwid, + rc = coalesce_paths(&vecs, newmp, refwwid, conf->force_reload, cmd); r = rc == CP_RETRY ? RTVL_RETRY : rc == CP_OK ? RTVL_OK : RTVL_FAIL; out: + if (r == RTVL_OK && + (cmd == CMD_LIST_SHORT || cmd == CMD_LIST_LONG || + cmd == CMD_CREATE) && + (VECTOR_SIZE(curmp) > 0 || VECTOR_SIZE(newmp) > 0) && + !check_daemon()) + condlog(2, "Warning: multipath devices exist, but multipathd service is not running"); + if (refwwid) FREE(refwwid); free_multipathvec(curmp, KEEP_PATHS); vecs.mpvec = NULL; + free_multipathvec(newmp, KEEP_PATHS); free_pathvec(pathvec, FREE_PATHS); vecs.pathvec = NULL; -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [dm-devel] [PATCH v2 3/6] multipath: print warning if multipathd is not running. 2021-09-02 21:57 ` [dm-devel] [PATCH v2 3/6] multipath: print warning if multipathd is not running Benjamin Marzinski @ 2021-09-03 8:20 ` Martin Wilck 0 siblings, 0 replies; 11+ messages in thread From: Martin Wilck @ 2021-09-03 8:20 UTC (permalink / raw) To: bmarzins, christophe.varoqui; +Cc: dm-devel On Thu, 2021-09-02 at 16:57 -0500, Benjamin Marzinski wrote: > If multipath notices that multipath devices exist or were created, and > multipathd is not running, it now prints a warning message, so users > are > notified of the issue. > > Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> Reviewed-by: Martin Wilck <mwilck@suse.com> -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply [flat|nested] 11+ messages in thread
* [dm-devel] [PATCH v2 4/6] libmultipath: remove unneeded code in coalesce_paths 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski ` (2 preceding siblings ...) 2021-09-02 21:57 ` [dm-devel] [PATCH v2 3/6] multipath: print warning if multipathd is not running Benjamin Marzinski @ 2021-09-02 21:57 ` Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 5/6] libmultipath: deal with dynamic PTHREAD_STACK_MIN Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() Benjamin Marzinski 5 siblings, 0 replies; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck The code at the end of coalesce_paths() removes a multipath device that was just created/reloaded, if none of its path devices have pp->dev set. This code is very old, and no longer has any actual effect. Multipath devices no longer get placed in pathvec without having pp->dev set. Even if they could, there's no point in creating/reloading the device and then immediately removing it. Reviewed-by: Martin Wilck <mwilck@suse.com> Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- libmultipath/configure.c | 46 ---------------------------------------- 1 file changed, 46 deletions(-) diff --git a/libmultipath/configure.c b/libmultipath/configure.c index 2f417914..3f241c5c 100644 --- a/libmultipath/configure.c +++ b/libmultipath/configure.c @@ -1060,28 +1060,6 @@ int domap(struct multipath *mpp, char *params, int is_daemon) return DOMAP_FAIL; } -static int -deadmap (struct multipath * mpp) -{ - int i, j; - struct pathgroup * pgp; - struct path * pp; - - if (!mpp->pg) - return 1; - - vector_foreach_slot (mpp->pg, pgp, i) { - if (!pgp->paths) - continue; - - vector_foreach_slot (pgp->paths, pp, j) - if (strlen(pp->dev)) - return 0; /* alive */ - } - - return 1; /* dead */ -} - extern int check_daemon(void) { @@ -1313,30 +1291,6 @@ int coalesce_paths (struct vectors *vecs, vector mpvec, char *refwwid, remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); } - /* - * Flush maps with only dead paths (ie not in sysfs) - * Keep maps with only failed paths - */ - if (mpvec) { - vector_foreach_slot (newmp, mpp, i) { - char alias[WWID_SIZE]; - - if (!deadmap(mpp)) - continue; - - strlcpy(alias, mpp->alias, WWID_SIZE); - - vector_del_slot(newmp, i); - i--; - remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); - - if (dm_flush_map(alias)) - condlog(2, "%s: remove failed (dead)", - alias); - else - condlog(2, "%s: remove (dead)", alias); - } - } ret = CP_OK; out: free(size_mismatch_seen); -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply related [flat|nested] 11+ messages in thread
* [dm-devel] [PATCH v2 5/6] libmultipath: deal with dynamic PTHREAD_STACK_MIN 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski ` (3 preceding siblings ...) 2021-09-02 21:57 ` [dm-devel] [PATCH v2 4/6] libmultipath: remove unneeded code in coalesce_paths Benjamin Marzinski @ 2021-09-02 21:57 ` Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() Benjamin Marzinski 5 siblings, 0 replies; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck Starting in glibc-2.34 (commit 5d98a7da), PTHREAD_STACK_MIN is defined as sysconf(_SC_THREAD_STACK_MIN) if _GNU_SOURCE is defined. sysconf() returns a long and can, at least in theory, return -1. This change causes compilation to fail in setup_thread_attr() due to a comparision with different signedness, since stacksize is a size_t. Reviewed-by: Martin Wilck <mwilck@suse.com> Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- libmultipath/util.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/libmultipath/util.c b/libmultipath/util.c index e2fafe85..ea858409 100644 --- a/libmultipath/util.c +++ b/libmultipath/util.c @@ -223,8 +223,8 @@ setup_thread_attr(pthread_attr_t *attr, size_t stacksize, int detached) ret = pthread_attr_init(attr); assert(ret == 0); - if (stacksize < PTHREAD_STACK_MIN) - stacksize = PTHREAD_STACK_MIN; + if (PTHREAD_STACK_MIN > 0 && stacksize < (size_t)PTHREAD_STACK_MIN) + stacksize = (size_t)PTHREAD_STACK_MIN; ret = pthread_attr_setstacksize(attr, stacksize); assert(ret == 0); if (detached) { -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply related [flat|nested] 11+ messages in thread
* [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski ` (4 preceding siblings ...) 2021-09-02 21:57 ` [dm-devel] [PATCH v2 5/6] libmultipath: deal with dynamic PTHREAD_STACK_MIN Benjamin Marzinski @ 2021-09-02 21:57 ` Benjamin Marzinski 2021-09-03 9:28 ` Martin Wilck 5 siblings, 1 reply; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-02 21:57 UTC (permalink / raw) To: Christophe Varoqui; +Cc: device-mapper development, Martin Wilck When remove_map() is called, if the multipath device is in a mpvec, it must be removed from it, because the device will be freed. Now that the mpvec is passed as a separate parameter to remove_map(), the purge_vec parameter is redundant. It was only used by coalesce_paths(), since the multipath device isn't on any vector when remove_map() is called there. Instead, remove_map() can just be called with a NULL mpvec, when there is no mpvec to remove the device from. remove_map_by_alias() also has a redundant purge_vec parameter. Since it only removes a map if it finds in on vec->mpvec, calling it with KEEP_VEC would be a bug, since it would leave a pointer to the freed device in the vector. Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- libmpathpersist/mpath_persist.c | 2 +- libmultipath/configure.c | 13 ++++++------- libmultipath/structs_vec.c | 12 ++++++------ libmultipath/structs_vec.h | 11 ++--------- multipath/main.c | 4 ++-- multipathd/dmevents.c | 2 +- multipathd/main.c | 12 ++++++------ 7 files changed, 24 insertions(+), 32 deletions(-) diff --git a/libmpathpersist/mpath_persist.c b/libmpathpersist/mpath_persist.c index 26710e79..803a2a28 100644 --- a/libmpathpersist/mpath_persist.c +++ b/libmpathpersist/mpath_persist.c @@ -410,7 +410,7 @@ get_mpvec (vector curmp, vector pathvec, char * refwwid) if (update_multipath_table(mpp, pathvec, DI_CHECKER) != DMP_OK || update_mpp_paths(mpp, pathvec)) { condlog(1, "error parsing map %s", mpp->wwid); - remove_map(mpp, pathvec, curmp, PURGE_VEC); + remove_map(mpp, pathvec, curmp); i--; } else extract_hwe_from_path(mpp); diff --git a/libmultipath/configure.c b/libmultipath/configure.c index 3f241c5c..7edb355b 100644 --- a/libmultipath/configure.c +++ b/libmultipath/configure.c @@ -1198,7 +1198,7 @@ int coalesce_paths (struct vectors *vecs, vector mpvec, char *refwwid, if (!mpp->paths) { condlog(0, "%s: skip coalesce (no paths)", mpp->alias); - remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); + remove_map(mpp, vecs->pathvec, NULL); continue; } @@ -1226,7 +1226,7 @@ int coalesce_paths (struct vectors *vecs, vector mpvec, char *refwwid, verify_paths(mpp); if (setup_map(mpp, ¶ms, vecs)) { - remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); + remove_map(mpp, vecs->pathvec, NULL); continue; } @@ -1248,7 +1248,7 @@ int coalesce_paths (struct vectors *vecs, vector mpvec, char *refwwid, condlog(2, "%s: %s map", mpp->alias, (mpp->action == ACT_CREATE)? "ignoring" : "removing"); - remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); + remove_map(mpp, vecs->pathvec, NULL); continue; } else /* if (r == DOMAP_RETRY && !is_daemon) */ { ret = CP_RETRY; @@ -1257,7 +1257,7 @@ int coalesce_paths (struct vectors *vecs, vector mpvec, char *refwwid, } if (r == DOMAP_DRY) { if (!vector_alloc_slot(newmp)) { - remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); + remove_map(mpp, vecs->pathvec, NULL); goto out; } vector_set_slot(newmp, mpp); @@ -1282,14 +1282,13 @@ int coalesce_paths (struct vectors *vecs, vector mpvec, char *refwwid, if (mpp->action != ACT_REJECT) { if (!vector_alloc_slot(newmp)) { - remove_map(mpp, vecs->pathvec, vecs->mpvec, KEEP_VEC); + remove_map(mpp, vecs->pathvec, NULL); goto out; } vector_set_slot(newmp, mpp); } else - remove_map(mpp, vecs->pathvec, vecs->mpvec, - KEEP_VEC); + remove_map(mpp, vecs->pathvec, NULL); } ret = CP_OK; out: diff --git a/libmultipath/structs_vec.c b/libmultipath/structs_vec.c index 24d6fd24..85d97ac1 100644 --- a/libmultipath/structs_vec.c +++ b/libmultipath/structs_vec.c @@ -334,7 +334,7 @@ void set_path_removed(struct path *pp) } void -remove_map(struct multipath *mpp, vector pathvec, vector mpvec, int purge_vec) +remove_map(struct multipath *mpp, vector pathvec, vector mpvec) { int i; @@ -343,7 +343,7 @@ remove_map(struct multipath *mpp, vector pathvec, vector mpvec, int purge_vec) */ orphan_paths(pathvec, mpp, "map removed internally"); - if (purge_vec && + if (mpvec && (i = find_slot(mpvec, (void *)mpp)) != -1) vector_del_slot(mpvec, i); @@ -354,12 +354,12 @@ remove_map(struct multipath *mpp, vector pathvec, vector mpvec, int purge_vec) } void -remove_map_by_alias(const char *alias, struct vectors * vecs, int purge_vec) +remove_map_by_alias(const char *alias, struct vectors * vecs) { struct multipath * mpp = find_mp_by_alias(vecs->mpvec, alias); if (mpp) { condlog(2, "%s: removing map by alias", alias); - remove_map(mpp, vecs->pathvec, vecs->mpvec, purge_vec); + remove_map(mpp, vecs->pathvec, vecs->mpvec); } } @@ -373,7 +373,7 @@ remove_maps(struct vectors * vecs) return; vector_foreach_slot (vecs->mpvec, mpp, i) { - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); i--; } @@ -704,7 +704,7 @@ struct multipath *add_map_with_path(struct vectors *vecs, struct path *pp, return mpp; out: - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); return NULL; } diff --git a/libmultipath/structs_vec.h b/libmultipath/structs_vec.h index ee2b723b..29ede454 100644 --- a/libmultipath/structs_vec.h +++ b/libmultipath/structs_vec.h @@ -27,15 +27,8 @@ int update_mpp_paths(struct multipath * mpp, vector pathvec); int update_multipath_strings (struct multipath *mpp, vector pathvec); void extract_hwe_from_path(struct multipath * mpp); -enum { - KEEP_VEC, - PURGE_VEC, -}; - -void remove_map (struct multipath *mpp, vector pathvec, vector mpvec, - int purge_vec); -void remove_map_by_alias(const char *alias, struct vectors * vecs, - int purge_vec); +void remove_map (struct multipath *mpp, vector pathvec, vector mpvec); +void remove_map_by_alias(const char *alias, struct vectors * vecs); void remove_maps (struct vectors * vecs); void sync_map_state (struct multipath *); diff --git a/multipath/main.c b/multipath/main.c index 6c70760f..65ece830 100644 --- a/multipath/main.c +++ b/multipath/main.c @@ -191,14 +191,14 @@ get_dm_mpvec (enum mpath_cmds cmd, vector curmp, vector pathvec, char * refwwid) if (refwwid && strlen(refwwid) && strncmp(mpp->wwid, refwwid, WWID_SIZE)) { condlog(3, "skip map %s: out of scope", mpp->alias); - remove_map(mpp, pathvec, curmp, PURGE_VEC); + remove_map(mpp, pathvec, curmp); i--; continue; } if (update_multipath_table(mpp, pathvec, flags) != DMP_OK) { condlog(1, "error parsing map %s", mpp->wwid); - remove_map(mpp, pathvec, curmp, PURGE_VEC); + remove_map(mpp, pathvec, curmp); i--; continue; } diff --git a/multipathd/dmevents.c b/multipathd/dmevents.c index f52f5970..f035ee7f 100644 --- a/multipathd/dmevents.c +++ b/multipathd/dmevents.c @@ -359,7 +359,7 @@ static int dmevent_loop (void) pthread_testcancel(); r = 0; if (curr_dev.action == EVENT_REMOVE) - remove_map_by_alias(curr_dev.name, waiter->vecs, PURGE_VEC); + remove_map_by_alias(curr_dev.name, waiter->vecs); else r = update_multipath(waiter->vecs, curr_dev.name, 1); pthread_cleanup_pop(1); diff --git a/multipathd/main.c b/multipathd/main.c index ae8272e3..3aff241d 100644 --- a/multipathd/main.c +++ b/multipathd/main.c @@ -385,7 +385,7 @@ remove_map_and_stop_waiter(struct multipath *mpp, struct vectors *vecs) condlog(3, "%s: removing map from internal tables", mpp->alias); if (!poll_dmevents) stop_waiter_thread(mpp); - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); } static void @@ -518,7 +518,7 @@ retry: fail: if (new_map && (retries < 0 || wait_for_events(mpp, vecs))) { condlog(0, "%s: failed to create new map", mpp->alias); - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); return 1; } @@ -572,7 +572,7 @@ add_map_without_path (struct vectors *vecs, const char *alias) return mpp; out: - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); return NULL; } @@ -1162,7 +1162,7 @@ rescan: goto fail; fail_map: - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); fail: orphan_path(pp, "failed to add path"); return 1; @@ -1469,7 +1469,7 @@ map_discovery (struct vectors * vecs) vector_foreach_slot (vecs->mpvec, mpp, i) if (update_multipath_table(mpp, vecs->pathvec, 0) != DMP_OK) { - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); i--; } @@ -2706,7 +2706,7 @@ configure (struct vectors * vecs) */ vector_foreach_slot(vecs->mpvec, mpp, i) { if (wait_for_events(mpp, vecs)) { - remove_map(mpp, vecs->pathvec, vecs->mpvec, PURGE_VEC); + remove_map(mpp, vecs->pathvec, vecs->mpvec); i--; continue; } -- 2.17.2 -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() 2021-09-02 21:57 ` [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() Benjamin Marzinski @ 2021-09-03 9:28 ` Martin Wilck 2021-09-03 15:22 ` Martin Wilck 0 siblings, 1 reply; 11+ messages in thread From: Martin Wilck @ 2021-09-03 9:28 UTC (permalink / raw) To: bmarzins, christophe.varoqui; +Cc: dm-devel On Thu, 2021-09-02 at 16:57 -0500, Benjamin Marzinski wrote: > When remove_map() is called, if the multipath device is in a mpvec, > it > must be removed from it, because the device will be freed. Now that > the > mpvec is passed as a separate parameter to remove_map(), the > purge_vec > parameter is redundant. It was only used by coalesce_paths(), since > the > multipath device isn't on any vector when remove_map() is called > there. > Instead, remove_map() can just be called with a NULL mpvec, when > there > is no mpvec to remove the device from. > > remove_map_by_alias() also has a redundant purge_vec parameter. > Since > it only removes a map if it finds in on vec->mpvec, calling it with > KEEP_VEC would be a bug, since it would leave a pointer to the freed > device in the vector. > > Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> Good catch. But we need to adapt libmultipath.version. I'll send a fix to be added on top. Reviewed-by: Martin Wilck <mwilck@suse.com> -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() 2021-09-03 9:28 ` Martin Wilck @ 2021-09-03 15:22 ` Martin Wilck 2021-09-03 17:17 ` Benjamin Marzinski 0 siblings, 1 reply; 11+ messages in thread From: Martin Wilck @ 2021-09-03 15:22 UTC (permalink / raw) To: bmarzins, christophe.varoqui; +Cc: dm-devel On Fri, 2021-09-03 at 11:28 +0200, Martin Wilck wrote: > On Thu, 2021-09-02 at 16:57 -0500, Benjamin Marzinski wrote: > > When remove_map() is called, if the multipath device is in a mpvec, > > it > > must be removed from it, because the device will be freed. Now that > > the > > mpvec is passed as a separate parameter to remove_map(), the > > purge_vec > > parameter is redundant. It was only used by coalesce_paths(), > > since > > the > > multipath device isn't on any vector when remove_map() is called > > there. > > Instead, remove_map() can just be called with a NULL mpvec, when > > there > > is no mpvec to remove the device from. > > > > remove_map_by_alias() also has a redundant purge_vec parameter. > > Since > > it only removes a map if it finds in on vec->mpvec, calling it with > > KEEP_VEC would be a bug, since it would leave a pointer to the > > freed > > device in the vector. > > > > Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> > > Good catch. But we need to adapt libmultipath.version. I'll send a > fix > to be added on top. > > Reviewed-by: Martin Wilck <mwilck@suse.com> This breaks the directio test, too. We need a test patch on top. I'm going to post it. Martin -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() 2021-09-03 15:22 ` Martin Wilck @ 2021-09-03 17:17 ` Benjamin Marzinski 0 siblings, 0 replies; 11+ messages in thread From: Benjamin Marzinski @ 2021-09-03 17:17 UTC (permalink / raw) To: Martin Wilck; +Cc: dm-devel On Fri, Sep 03, 2021 at 03:22:45PM +0000, Martin Wilck wrote: > On Fri, 2021-09-03 at 11:28 +0200, Martin Wilck wrote: > > On Thu, 2021-09-02 at 16:57 -0500, Benjamin Marzinski wrote: > > > When remove_map() is called, if the multipath device is in a mpvec, > > > it > > > must be removed from it, because the device will be freed. Now that > > > the > > > mpvec is passed as a separate parameter to remove_map(), the > > > purge_vec > > > parameter is redundant. It was only used by coalesce_paths(), > > > since > > > the > > > multipath device isn't on any vector when remove_map() is called > > > there. > > > Instead, remove_map() can just be called with a NULL mpvec, when > > > there > > > is no mpvec to remove the device from. > > > > > > remove_map_by_alias() also has a redundant purge_vec parameter. > > > Since > > > it only removes a map if it finds in on vec->mpvec, calling it with > > > KEEP_VEC would be a bug, since it would leave a pointer to the > > > freed > > > device in the vector. > > > > > > Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> > > > > Good catch. But we need to adapt libmultipath.version. I'll send a > > fix > > to be added on top. > > > > Reviewed-by: Martin Wilck <mwilck@suse.com> > > This breaks the directio test, too. We need a test patch on top. > I'm going to post it. Oops. Thanks. -Ben > > Martin -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel ^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2021-09-03 17:18 UTC | newest] Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2021-09-02 21:57 [dm-devel] [PATCH v2 0/6] Mulitpath: miscellaneous patches Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 1/6] multipath.conf: fix typo in ghost_delay description Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 2/6] mpathpersist: fail commands when no usable paths exist Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 3/6] multipath: print warning if multipathd is not running Benjamin Marzinski 2021-09-03 8:20 ` Martin Wilck 2021-09-02 21:57 ` [dm-devel] [PATCH v2 4/6] libmultipath: remove unneeded code in coalesce_paths Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 5/6] libmultipath: deal with dynamic PTHREAD_STACK_MIN Benjamin Marzinski 2021-09-02 21:57 ` [dm-devel] [PATCH v2 6/6] libmultipath: drop unnecessary parameter from remove_map() Benjamin Marzinski 2021-09-03 9:28 ` Martin Wilck 2021-09-03 15:22 ` Martin Wilck 2021-09-03 17:17 ` Benjamin Marzinski
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).