* [PATCH v12 0/8] Xen VMware tools support
@ 2015-06-27 23:27 Don Slutz
2015-06-27 23:27 ` [PATCH v12 1/8] tools: Add vga=vmware Don Slutz
` (8 more replies)
0 siblings, 9 replies; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Tim Deegan,
George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
Changes v11 to v12:
Rebased on staging.
Jan Beulich (xen: Add ring 3 vmware_port support)
As there don't seem to be enough convincing arguments for this
to be worthwhile, I'm going to drop this...
Dropped the 2 ring3 patches.
Ian Campbell (tools: Add vmware_hwver support):
s/come/comes/
Done
I'm not sure this hunk has anything to do with this patch, nor
what the semantic difference between the old and new text is
supposed to be.
Dropped comment change.
Ian Campbell (multiple):
In v11 this seems to have morphed into only
LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE being provided, which
is clearly not an appropriate umbrella #define.
"#define LIBXL_HAVE_CREATEINFO_VMWARE 1"
Lets just have a single one of these indicating support for
vmware, it should be added at the end of the series after all
the baseline vmware functionality is in place. I think that
means hwver, vga=vmware and this port stuff.
Make (tools: Add vga=vmware) no longer independent.
Change the #define to "LIBXL_HAVE_VMWARE"
Jan Beulich (xen: Add vmware_port support):
Surrounding code avoiding the use of "break" makes the result
look rather inconsistent. Please move this up immediately after
the XSM check, or drop the "break".
Moved it up.
George Dunlap (Add IOREQ_TYPE_VMWARE_PORT):
Sorry for coming a bit late to this party. On a high level I
think this is good, but there doesn't seem to be anything in
here in particular that is vmware-specific. Would it make more
sense to give this a more generic name, and have it include all
of the general-purpose registers?
Did not change at this time because QEMU would need changes 1st.
George Dunlap (Add xentrace to vmware_port):
Do you need to log edi as well? It looks like it's not used.
I think it makes sense to indicate that edi can be used and
may in the future.
And do you need to log all the registers here? It seems like
port + regs->_ecx would be enough to tell you why it got
ignored.
Adjusted to log just port and eax (ecx was NOT tested for).
Changes v10 to v11:
Andrew Cooper & Ian Campbell (#1 "tools: Add vga=vmware"):
Nack. Qemu-trad is currently has remote code execution vulnerabilities.
Dropped support for Qemu-trad.
Also changed later patchs to not need this one.
Andrew Cooper (#2 "xen: Add support for VMware cpuid leaves"):
Needs re-base.
Done
Adjust /* Disallow if vmware_hwver */
Done
Newline after break;
Done 2 places.
Allowed Reviewed-by: Andrew Cooper, if these changes are done.
Added Reviewed-by: Andrew Cooper.
Julien Grall (#2 "xen: Add support for VMware cpuid leaves"):
It would be worth to add an explicit vmware_hwver = 0 in the
libxl__arch_domain_prepare_config.
Done -- Note: Adds a tool change to this patch.
(#3 "tools: Add vmware_hwver support"):
Since Qemu-trad does not support vga=vmware,
Dropped "If non-zero then default VGA to VMware's VGA"
Andrew Cooper (#5 "xen: Add vmware_port support"):
You will not be getting here for a non HVM domain...
Dropped ASSERT(is_hvm_domain(currd))
Newline after break;
Done 6 places.
Allowed Reviewed-by: Andrew Cooper, if these changes are done.
Added Reviewed-by: Andrew Cooper.
(#7 "tools: Add vmware_port support"):
Since Qemu-trad does not support vga=vmware,
Dropped "If non-zero then default VGA to VMware's VGA"
Changes v9 to v10:
Split out LIBXL_VGA_INTERFACE_TYPE_VMWARE into it's own patch (#1)
that can stand alone. In the patch set because a later patch
depends on it.
Reworked to be based on:
commit a7511905fae7ba592c5bf63cd77d8ff78087d689
Author: Julien Grall <julien.grall@linaro.org>
Date: Wed Apr 1 17:21:41 2015 +0100
xen: Extend DOMCTL createdomain to support arch configuration
rebased onto:
commit e13013dbf1d5997915548a3b5f1c39594d8c1d7b
Author: Yang Hongyang <yanghy@cn.fujitsu.com>
Date: Thu May 14 16:55:18 2015 +0800
libxc/restore: add checkpointed flag to the restore context
Andrew Cooper (#2: "xen: Add support for VMware cpuid leaves"):
Did not add "Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>"
because of changes here to do things the new way.
Reword comment message to reflect new way.
Ian Campbell (#3 "tools: Add vmware_hwver support"):
LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE &
LIBXL_HAVE_BUILDINFO_HVM_VMWARE_HWVER are arriving together
a single umbrella could be used.
Since I split the LIBXL_VGA_INTERFACE_TYPE_VMWARE into
it's own patch, this is not longer true.
But I did use 1 for the 2 c_info changes.
Please use GCSPRINTF.
Done.
Remove vga=vmware from here.
Ian Campbell (#3 "tools: Add vmware_hwver support"):
For "Add IOREQ_TYPE_VMWARE_PORT"
With those fixed the tools/* bits are:
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Did not add Acked-by to "tools: Add vmware_hwver support"
because of the rework for using libxl_domain_create_info.
Andrew Cooper (#4: "vmware: Add VMware provided include file."):
Added "Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>"
Andrew Cooper (#5 "xen: Add vmware_port support"):
Probably better as EOPNOTSUPP, as it is a configuration problem.
Done.
vmport_ioport function looks as if it should be static.
Done.
Why is GETHZ the only one of these with a CPL check?
Please see thread for detail.
I would suggest putting vmport_register declaration in hvm.h ...
Done.
Jan Beulich (#5 "xen: Add vmware_port support"):
As indicated before, I don't think this is a good use case for a
domain creation flag.
Switch to the new config way.
struct domain *d => struct domain *currd
Done
Are you sure you don't want to zero the high halves of 64-bit ...
Comment added.
Then just have this handled into the default case.
Reworked new_eax handling.
is_hvm_domain(currd)
And - why here rather than before the switch() or even right at the
start of the function?
Moved to start.
With that, is it really correct that OUT updates the other registers
just like IN? If so, this deserves a comment, so that readers won't
think this is in error.
All done in comment at start.
Andrew Cooper (#6 "xen: Add ring 3 vmware_port support"):
>> This looks horribly invasive.
>>
>> Why are emulation changes needed? What is wrong with the normal
>> handling with a registered ioport handler?
> Because VMware made a bad way to provide a "hyper call". They decided to
> allow user access to this. So when a #GP fault should have been
> reported, they instead do the "hyper call".
>
Urgh - now I remember.
Right. In the case that vmport is active, we start intercepting #GP
faults and emulating access. That part of the patch looks ok.
However, the rest is very invasive to the emulation infrastructure.
Re-worked along this lines suggested.
Jan Beulich (#6 "xen: Add ring 3 vmware_port support"):
I hope that vmport_check will no longer be needed with the adjustments ...
> Since this is not an architecture feature and I do not expect any real
> CPUs to support this, I do not expect any other use. But I am happy
> to make it more generic.
Let's see how this ends up looking - the hook is probably indeed
bogus (from an architectural pov) no matter how you name it.
Last e-mail on thread, so no change.
Ian Campbell (#7 "tools: Add vmware_port support"):
If..." at the start of the sentence ...
Used Ian's reword.
Also, why is 7 special?
Attempted to better explain.
Paul Durrant & Jan Beulich (#8 "Add IOREQ_TYPE_VMWARE_PORT"):
Now that buf is no longer a bool, could ...
These literals should become an enum
Added an enum.
I don't think the invalidate type is needed.
Dropped.
IOREQ_TYPE_VMWARE_PORT as 3 is a re-use.
Switch to 9.
Code handling "case X86EMUL_UNHANDLEABLE:" in emulate.c
is unclear.
Re-worked to a version that Jan likes better.
Comment about "special' range of 1" is not clear.
Re-worded comments.
Ian Campbell (#9 "Add xentrace to vmware_port"):
Acked-by
Readded dropped traces.
Jan Beulich & Andrew Cooper (#9 "Add xentrace to vmware_port"):
Why is cmd in this patch?
Because the trace points use it.
Jan Beulich (#10 "test_x86_emulator.c: Add tests for #GP usage"):
Need more comments and simpler error checking.
Done.
Dropped un-needed new routines.
Andrew Cooper:
That is because you broke it adding a bool_t item.
Has now been dropped.
Changes v8 to v9:
Overview of changes:
s/vmware_hw/vmware_hwver/i
Switch to x86_emulator to handle #GP
New patch: Move MAX_INST_LEN into x86_emulate.h
Add QEMU usage, patch #8 "Add IOREQ_TYPE_VMWARE_PORT"
Split patch "xen: Add vmware_port support" into 2. 1st has same
name. New one is "xen: Add ring 3 vmware_port support".
Added 3 new patches about test_x86_emulator.
Jan Beulich (#2: "xen: Add support for VMware cpuid leaves"):
Change -EXDEV to EOPNOTSUPP.
Done.
adding another subdirectory: xen/arch/x86/hvm/vmware
Much will depend on the discussion of the subsequent patches.
TBD.
So for versions < 7 there's effectively no CPUID support at all?
Changed to check at entry.
The comment /* Params for VMware */ seems wrong...
Changed to /* emulated VMware Hardware Version */
Also please use d, not _d in #define is_vmware_domain()
Changed. Line is now > 80 characters, so split into 2.
Andrew Cooper (#3: "tools: Add vmware_hwver support"):
I assumed that s/vmware_hw/vmware_hwver/ is not a big enough
change to drop the Reviewed-by. Did a minor edit to the
commit message to add 7 to the list of values checked.
Jan Beulich (#4: "vmware: Add VMware provided include file"):
Either the description is wrong, or the patch is stale.
stale commit message -- fixed.
I'd say a file with a single comment line in it would suffice.
Done.
Jan Beulich (#5: "xen: Add vmware_port support"):
Can you explain why a HVM param isn't suitable here?
Issue with changing QEMU on the fly.
Andrew Cooper: My recommendation is still to use a creation flag
So no change.
Please move SVM's identical definition into ...
Did this as #1. No longer needed, but since the patch was ready
I have included it.
--Lots of questions about code that no long is part of this patch. --
With this, is handling other than 32-bit in/out really
meaningful/correct?
Added comment about this.
Since you can't get here for PV, I can't see what you need this.
Changed to an ASSERT.
Why version 4?
Added comment about this.
-- Several questions about register changes.
Re-coded to use new_eax and set *val to this.
Change to generealy use reg->_e..
These ei1/ei2 checks belong in the callers imo -
Moved.
the "port" function parameter isn't even checked
Add check for exact match.
If dropping the code is safe without also forbidding the
combination of nested and VMware emulation.
Added the forbidding the combination of nested and VMware.
Mostly do to the cases of the nested virtual code is the one
to handle VMware stuff if needed, not the root one. Also I am
having issues testing xen nested in xen and using hvm.
Changes v7 to v8:
Jan Beulich:
Coding changes to vmport_ioport. Things like:
- regs->rax = (uint32_t)~0ul;
+ regs->_eax = ~0u;
Andrew Cooper (#2: "tools: Add vmware_hwver support"):
Other than these two comments, the rest of the patch looks ok, so...
Added Reviewed-by after addressing the "Spurious whitepsace change".
and the wording in the new docs/misc/hypervisor-cpuid.markdown.
Changes v6 to v7:
summary of changes.
George Dunlap:
Any doc about this?
Added reference to:
https://sites.google.com/site/chitchatvmback/backdoor
Last updated: Feb. 2008
George Dunlap & Jan Beulich
Too much logging and tracing.
Dropped a lot of it. This includes vmport_debug=
Ian Campbell:
Any reason RPC code cannot be done in QEMU?
Not that I know of, so dropped all parts of RPC code.
Default handling of hvm.vga.kind bad.
Fixed.
Default of vmware_port should be based on vmware_hw.
Done.
Tim Deegan:
CPL check of GETHZ needs to be fixed somewhere.
Added check for CPL == 0 (assuming this is what VMware is
checking. Matches the testing.
Ian Campbell, Andrew Cooper, George Dunlap, Boris Ostrovsky,
& Jan Beulich
Various minor fixes.
Per patch notes:
#1 "xen: Add support for VMware cpuid leaves":
Prevent setting of HVM_PARAM_VIRIDIAN if HVM_PARAM_VMWARE_HW set.
#4 "xen: Add vmware_port support":
More on AMD in the commit message.
Switch to only change 32bit part of registers, what VMware
does.
#6 "Add xentrace to vmware_port":
Dropped some of the new traces.
Added HVMTRACE_ND7.
#7 "Add xen-hvm-param":
Was a later patch. Still optional.
Fixed formatting.
Adjust for drop of VMware RPC.
Comments on v3, v4, v5, v6:
George Dunlap:
Is there any reason not to merge 05/16 with 03/16?
The reason I have is that v3 03/16 only contains new files. 2
from VMware and 1 to allow use of the VMware files. I added
xen/arch/x86/hvm/vmware/includeCheck.h at the request of
Konrad Wilk.
This patch has many style issues and white space issues. So I
want it as a separate patch so as to be clear on what files do
not meet the coding style. And why and where they came from.
Changes v5 to v6:
Boris Ostrovsky & Jan Beulich
#4 "xen: Add vmware_port support":
#6 "xen: Convert vmware_port to xentrace usage":
There is an issue with reading instruction bytes more then once.
Dropped the attempt to use svm_nextrip_insn_length via
__get_instruction_length (added in v2). Just always look
at upto 15 bytes on AMD.
Changes v4 to v5:
Re tagged the optional patches.
Added debug=y build checking that vmx is defining
VM_EXIT_INTR_ERROR_CODE.
Boris Ostrovsky:
#1 "xen: Add support for VMware cpuid leaves":
Given how is_viridian and is_vmware are defined I think '||' is more
appropriate.
Fixed.
#4 "xen: Add is_vmware_port_enabled":
we should make sure that svm_vmexit_gp_intercept is not executed for
any other guest.
Added an ASSERT on is_vmware_port_enabled.
magic integers?
Added #define for them.
#6 "xen: Convert vmware_port to xentrace usage":
exitinfo1 is used twice.
Fixed.
#7 "tools: Convert vmware_port to xentrace usage":
'bytes = 0x%(2)d' or 'bytes = %(2)d' ?
Fixed.
#8 "xen: Add limited support of VMware's hyper-call rpc":
PV vs. HVM vs. PVH. So probably 'if(is_hvm_vcpu)'?
I see no reason to exclude PVH. Will change to has_hvm_container_vcpu
#11 "Add live migration of VMware's hyper-call":
You ASSERTed that vg->key_len is 1 so you may not need the 'if'.
That is a ASSERT(sizeof, not just ASSERT -- not changed.
Use real errno, not -1.
Fixed.
No ASSERT in vmport_load_domain_ctxt
Added.
Jan Beulich & Boris Ostrovsky:
#8 "xen: Add limited support of VMware's hyper-call rpc":
The names of all three functions are bogus.
removed static support routines.
Also changed in #1.
Andrew Cooper:
#2 "tools: Add vmware_hw support":
Anything looking for Xen according to the Xen cpuid instructions...
Adjusted doc to new wording.
#4 "xen: Add is_vmware_port_enabled":
I am fairly certain that you need some brackets here.
Added brackets.
Jan Beulich & Andrew Cooper:
#1 "xen: Add support for VMware cpuid leaves":
This hunk is unrelated, but is perhaps something better fixed.
Added to commit message.
include <xen/types.h> (IIRC) please.
Done.
At least 1 pair of brackets please, especially as the placement of
brackets affects the result of this particular calculation.
Switch to "1000000ull / APIC_BUS_CYCLE_NS"
Changes v3 to v4:
Ian Campbell:
Report on both viridian and vmware_hw set.
Added LIBXL_VGA_INTERFACE_TYPE_VMWARE (vga=vmware).
Andrew Cooper:
Add doc for hypervisor-cpuid.
Boris Ostrovsky:
Changing regs->error_code may not be a good idea.
Dropped this.
Jan Beulich & Boris Ostrovsky:
Only enable vmwxit for GP when vmware_port is set.
Done.
Changes v2 to v3:
Add optional unit test tools.
Re-worked split of changes.
Jan Beulich:
for #0:
I don't think you should be adding a new fine in hvm/ _and_ a new
subdirectory.
Moved all files to hvm/vmware that contain code.
for old #1 (now #1 & #2):
Is there really a point in enabling both Viridian and VMware extensions?
I still think so.
hvmloader change: This needs an explanation
Dropped as not need now.
Can you make vmware_hw similar to Viridian, returning success when
setting the value to what it already is.
Done.
You don't seem to be using sub_idx: ...
Dropped.
Extra changes...
Dropped.
for old #2 (now #3):
... these guards have the (theoretical at this point) risk of clashing
... the patch is obviously incomplete without this header...
Did not fix any of these issues. I will stick with this needs
to be a 2nd patch that changes the include files to better fit
in Xen coding. For now these files are in a sub directory
which is not part of the normal include search.
Moved the includeCheck.h file into this patch.
for old #3 (now #4, #5, #6, #7, #8, #9, #10, #11)
As I think was said on v1 already - this should be split into smaller
pieces ...
Done.
All this would very likely better go into a separate function placed in
vmport.c.
Moved most of the code into vmport.c or vmport_rpc.c.
In any event I'm rather uncomfortable about vmware_port getting
enabled unconditionally, ...
Added vmware_port (done in new patches #4, #5) as an xl.cfg
option.
You'll have to go through and fix coding style issues.
I think I have found all these, but since they do not stand out
for me, let me know of any left.
"MAKE_INSTR(IN," name is ambiguous.
Added all 4 opcodes for in and out that can access this port: INB_DX,
INL_DX, OUTB_DX, OUTL_DX.
A VMX-specific function shouldn't be named this way...
Added new common routine vmport_gp_check() that is called from
both vmx.c and svm.c which is where all the logic about checking
for IN ans OUT is done.
Also fixed naming and added static.
Ah, here we go (as to using HVM_DBG_LOG()): Isn't this _way_ too
fine grained?
I have reduced the number of bits used. Partialy by switching
some to xentrace (new patch #6 and #7).
Right, and zero is an indication that it wasn't found. Also I just
noticed there's a gdprintk() in that event, which for all other ...
Made the gdprintk() optional.
End of v3 changes.
This is a small part of the changes needed to allow running Linux
and windows (and others) guests that were built on VMware and run
run them unchanged on Xen.
This small part is the start of Xen support of VMware backdoor I/O
port which is how VMware tools (a standard addition installed on a
guest) communicates to the hypervisor.
I picked this subset to start with because it only has changes in
Xen.
Some of this code is already in QEMU and so KVM has some of this
already. QEMU supported backdoor commands include VMware mouse
support. A later patch set exists that links these changes, new
code and Xen changes to QEMU to provide VMware mouse support under
Xen. The important part is that VMware mouse is an absolute
position mouse and so network delays do not effect usage of the
virtual mouse.
For example from the guest:
[root@C63-min-tools ~]# vmtoolsd --cmd "info-get guestinfo.joejoel"
No value found
[root@C63-min-tools ~]# vmtoolsd --cmd "info-set guestinfo.joejoel short"
[root@C63-min-tools ~]# vmtoolsd --cmd "info-get guestinfo.joejoel"
short
[root@C63-min-tools ~]# vmtoolsd --cmd "info-set guestinfo.joejoel long222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222200000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000joel"
[root@C63-min-tools ~]# vmtoolsd --cmd "info-get guestinfo.key1"
data1
[root@C63-min-tools ~]# vmtoolsd --cmd "info-get guestinfo.key2"
No value found
[root@C63-min-tools ~]# vmtoolsd --cmd "info-get guestinfo.key2"
data2
[root@C63-min-tools ~]#
Most of this code has been reverse engineered by looking at
source code for Linux and open VMware tools.
http://open-vm-tools.sourceforge.net
changes RFC to v2:
Jan Beulich:
Add xen/arch/x86/hvm/vmware.c for cpuid_vmware_leaves
Fewer patches
Andrew Cooper:
use the proper constant for apic_khz
Follow 839b966e3f587bbb1a0d954230fb3904330dccb6 style changes.
Changed HVM_PARAM_VMWARE_HW to write once (make is_vmware_domain()
more static).
Dropped vmport status stuff.
Added checks for xzalloc() having failed.
You should include backdoor_def.h ...
Every thing I tried did not work better. So I did not
change VMPORT_PORT and BDOOR_PORT being the same value.
I did not try and adjust VMware's include file backdoor_def.h
to working in other xen source files.
Switching to s_time_t is not valid. get_sec() is defined:
unsigned long get_sec(void);
and so my uses of it should be using unsigned long. However
since that is not a fixed width type, I used the uint64_t
data type which is almost the same, but does allow the 32 bit
build of libxc, libxl to do the correct thing.
Konrad Rzeszutek Wilk:
Please don't include the address. It should be, etc
about the Vmware provided include files.
I went with no changes to these files. Even if the files should
be changed to match xen coding style, etc I still feel that the
original ones should be added via a patch, and then adjusted in a
2nd patch.
Can you use XenBus?
I would say no. XenBus (and XenStore) is about domain to domain
communication. This is about VMware's hyper-call and providing
access to VMware's guest info very low speed access.
Olaf Hering:
Dropped changing of bios-strings. Still needs some documentation
about this may be needed to do in a tool stack or set of commands.
Boris Ostrovsky:
Use svm_nextrip_insn_length()
Looks like __get_instruction_length() does this, so switched to
__get_instruction_length().
RFC:
See
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1009458
for info on detecting VMware.
Linux does not follow this exactly. It checks for CPUID 1st. If
that fails, it checks for SMBIOS containing "VMware" (not VMware- or
VMW).
So this patch set provides:
SMBIOS -- Add string VMware-
CPUID -- Add VMware's CPUID (Note: currently HyperV (viridian support) breaks this check.)
Add the magic VMware port
Allow VMware tools poweroff and reboot
Enable access to VMware's guest info
Provide the VMware tools build number
Don Slutz (8):
tools: Add vga=vmware
xen: Add support for VMware cpuid leaves
tools: Add vmware_hwver support
vmware: Add VMware provided include file.
xen: Add vmware_port support
tools: Add vmware_port support
Add IOREQ_TYPE_VMWARE_PORT
Add xentrace to vmware_port
docs/man/xl.cfg.pod.5 | 36 +++++-
tools/libxc/xc_hvm_build_x86.c | 5 +-
tools/libxl/libxl.h | 5 +
tools/libxl/libxl_create.c | 13 ++-
tools/libxl/libxl_dm.c | 11 ++
tools/libxl/libxl_types.idl | 3 +
tools/libxl/libxl_x86.c | 5 +-
tools/libxl/xl_cmdimpl.c | 5 +
tools/xentrace/formats | 5 +
xen/arch/x86/domain.c | 3 +
xen/arch/x86/hvm/Makefile | 1 +
xen/arch/x86/hvm/emulate.c | 83 +++++++++++---
xen/arch/x86/hvm/hvm.c | 202 +++++++++++++++++++++++++++++----
xen/arch/x86/hvm/io.c | 19 ++++
xen/arch/x86/hvm/vmware/Makefile | 2 +
xen/arch/x86/hvm/vmware/backdoor_def.h | 167 +++++++++++++++++++++++++++
xen/arch/x86/hvm/vmware/cpuid.c | 77 +++++++++++++
xen/arch/x86/hvm/vmware/includeCheck.h | 1 +
xen/arch/x86/hvm/vmware/vmport.c | 170 +++++++++++++++++++++++++++
xen/arch/x86/traps.c | 8 +-
xen/include/asm-x86/hvm/domain.h | 9 +-
xen/include/asm-x86/hvm/hvm.h | 10 ++
xen/include/asm-x86/hvm/trace.h | 22 ++++
xen/include/asm-x86/hvm/vmware.h | 33 ++++++
xen/include/public/arch-x86/xen.h | 8 +-
xen/include/public/hvm/hvm_op.h | 5 +
xen/include/public/hvm/ioreq.h | 17 +++
xen/include/public/hvm/params.h | 4 +-
xen/include/public/trace.h | 3 +
29 files changed, 886 insertions(+), 46 deletions(-)
create mode 100644 xen/arch/x86/hvm/vmware/Makefile
create mode 100644 xen/arch/x86/hvm/vmware/backdoor_def.h
create mode 100644 xen/arch/x86/hvm/vmware/cpuid.c
create mode 100644 xen/arch/x86/hvm/vmware/includeCheck.h
create mode 100644 xen/arch/x86/hvm/vmware/vmport.c
create mode 100644 xen/include/asm-x86/hvm/vmware.h
--
1.8.3.1
^ permalink raw reply [flat|nested] 29+ messages in thread
* [PATCH v12 1/8] tools: Add vga=vmware
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-07-07 15:54 ` Ian Campbell
2015-06-27 23:27 ` [PATCH v12 2/8] xen: Add support for VMware cpuid leaves Don Slutz
` (7 subsequent siblings)
8 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
This allows use of QEMU's VMware emulated video card
NOTE: vga=vmware is not supported by device_model_version=qemu-xen-traditional
Signed-off-by: Don Slutz <dslutz@verizon.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v12:
Dropped LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE
This means that the later patch that defines LIBXL_HAVE_VMWARE
is now also required.
v11:
Dropped support for Qemu-trad.
Also changed later patchs to not need this one.
v10: New at v10.
Was part of "tools: Add vmware_hwver support"
docs/man/xl.cfg.pod.5 | 4 +++-
tools/libxl/libxl_dm.c | 9 +++++++++
tools/libxl/libxl_types.idl | 1 +
tools/libxl/xl_cmdimpl.c | 2 ++
4 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
index a3e0e2e..84078f6 100644
--- a/docs/man/xl.cfg.pod.5
+++ b/docs/man/xl.cfg.pod.5
@@ -1389,7 +1389,7 @@ This option is deprecated, use vga="stdvga" instead.
=item B<vga="STRING">
-Selects the emulated video card (none|stdvga|cirrus|qxl).
+Selects the emulated video card (none|stdvga|cirrus|qxl|vmware).
The default is cirrus.
In general, QXL should work with the Spice remote display protocol
@@ -1397,6 +1397,8 @@ for acceleration, and QXL driver is necessary in guest in this case.
QXL can also work with the VNC protocol, but it will be like a standard
VGA without acceleration.
+NOTE: vmware is not supported on B<device_model_version = "qemu-xen-traditional">
+
=item B<vnc=BOOLEAN>
Allow access to the display via the VNC protocol. This enables the
diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
index 6b8b3ac..975996d 100644
--- a/tools/libxl/libxl_dm.c
+++ b/tools/libxl/libxl_dm.c
@@ -256,6 +256,10 @@ static int libxl__build_device_model_args_old(libxl__gc *gc,
case LIBXL_VGA_INTERFACE_TYPE_NONE:
flexarray_append_pair(dm_args, "-vga", "none");
break;
+ case LIBXL_VGA_INTERFACE_TYPE_VMWARE:
+ LOG(ERROR, "vga=vmware is not supported by "
+ "qemu-xen-traditional");
+ return ERROR_INVAL;
case LIBXL_VGA_INTERFACE_TYPE_QXL:
break;
}
@@ -657,6 +661,11 @@ static int libxl__build_device_model_args_new(libxl__gc *gc,
GCSPRINTF("qxl-vga,vram_size_mb=%"PRIu64",ram_size_mb=%"PRIu64,
(b_info->video_memkb/2/1024), (b_info->video_memkb/2/1024) ) );
break;
+ case LIBXL_VGA_INTERFACE_TYPE_VMWARE:
+ flexarray_append_pair(dm_args, "-device",
+ GCSPRINTF("vmware-svga,vgamem_mb=%d",
+ libxl__sizekb_to_mb(b_info->video_memkb)));
+ break;
}
if (b_info->u.hvm.boot) {
diff --git a/tools/libxl/libxl_types.idl b/tools/libxl/libxl_types.idl
index 736c4e4..534e7f0 100644
--- a/tools/libxl/libxl_types.idl
+++ b/tools/libxl/libxl_types.idl
@@ -187,6 +187,7 @@ libxl_vga_interface_type = Enumeration("vga_interface_type", [
(2, "STD"),
(3, "NONE"),
(4, "QXL"),
+ (5, "VMWARE"),
], init_val = "LIBXL_VGA_INTERFACE_TYPE_CIRRUS")
libxl_vendor_device = Enumeration("vendor_device", [
diff --git a/tools/libxl/xl_cmdimpl.c b/tools/libxl/xl_cmdimpl.c
index 568fd21..efb3651 100644
--- a/tools/libxl/xl_cmdimpl.c
+++ b/tools/libxl/xl_cmdimpl.c
@@ -2142,6 +2142,8 @@ skip_vfb:
b_info->u.hvm.vga.kind = LIBXL_VGA_INTERFACE_TYPE_NONE;
} else if (!strcmp(buf, "qxl")) {
b_info->u.hvm.vga.kind = LIBXL_VGA_INTERFACE_TYPE_QXL;
+ } else if (!strcmp(buf, "vmware")) {
+ b_info->u.hvm.vga.kind = LIBXL_VGA_INTERFACE_TYPE_VMWARE;
} else {
fprintf(stderr, "Unknown vga \"%s\" specified\n", buf);
exit(1);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 2/8] xen: Add support for VMware cpuid leaves
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
2015-06-27 23:27 ` [PATCH v12 1/8] tools: Add vga=vmware Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-07-01 20:01 ` Konrad Rzeszutek Wilk
2015-07-07 15:56 ` Ian Campbell
2015-06-27 23:27 ` [PATCH v12 3/8] tools: Add vmware_hwver support Don Slutz
` (6 subsequent siblings)
8 siblings, 2 replies; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
This is done by adding xen_arch_domainconfig vmware_hw. It is set to
the VMware virtual hardware version.
Currently 0, 3-4, 6-11 are good values. However the
code only checks for == 0 or != 0 or >= 7.
If non-zero then
Return VMware's cpuid leaves. If >= 7 return data, else
return 0.
The support of hypervisor cpuid leaves has not been agreed to.
MicroSoft Hyper-V (AKA viridian) currently must be at 0x40000000.
VMware currently must be at 0x40000000.
KVM currently must be at 0x40000000 (from Seabios).
Xen can be found at the first otherwise unused 0x100 aligned
offset between 0x40000000 and 0x40010000.
http://download.microsoft.com/download/F/B/0/FB0D01A3-8E3A-4F5F-AA59-08C8026D3B8A/requirements-for-implementing-microsoft-hypervisor-interface.docx
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1009458
http://lwn.net/Articles/301888/
Attempted to get this cleaned up.
So based on this, I picked the order:
Xen at 0x40000000 or
Viridian or VMware at 0x40000000 and Xen at 0x40000100
If both Viridian and VMware selected, report an error.
Since I need to change xen/arch/x86/hvm/Makefile; also add
a newline at end of file.
Signed-off-by: Don Slutz <dslutz@verizon.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v12:
No change
v11:
Adjust /* Disallow if vmware_hwver */
Newline after break;
Added Reviewed-by: Andrew Cooper.
It would be worth to add an explicit vmware_hwver = 0 in the
libxl__arch_domain_prepare_config.
Note: Adds a tool change to this patch.
v10:
Did not add "Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>"
because of changes here to do things the new way.
Reword comment message to reflect new way.
v9:
s/vmware_hw/vmware_hwver/i
Change -EXDEV to EOPNOTSUPP.
Done.
adding another subdirectory: xen/arch/x86/hvm/vmware
Much will depend on the discussion of the subsequent patches.
TBD.
So for versions < 7 there's effectively no CPUID support at all?
Changed to check at entry.
The comment /* Params for VMware */ seems wrong...
Changed to /* emulated VMware Hardware Version */
Also please use d, not _d in #define is_vmware_domain()
Changed. Line is now > 80 characters, so split into 2.
v7:
Prevent setting of HVM_PARAM_VIRIDIAN if HVM_PARAM_VMWARE_HW set.
v5:
Given how is_viridian and is_vmware are defined I think '||' is more
appropriate.
Fixed.
The names of all three functions are bogus.
removed static support routines.
This hunk is unrelated, but is perhaps something better fixed.
Added to commit message.
include <xen/types.h> (IIRC) please.
Done.
At least 1 pair of brackets please, especially as the placement of
brackets affects the result of this particular calculation.
Switch to "1000000ull / APIC_BUS_CYCLE_NS"
tools/libxl/libxl_x86.c | 4 +-
xen/arch/x86/domain.c | 1 +
xen/arch/x86/hvm/Makefile | 1 +
xen/arch/x86/hvm/hvm.c | 11 ++++++
xen/arch/x86/hvm/vmware/Makefile | 1 +
xen/arch/x86/hvm/vmware/cpuid.c | 77 +++++++++++++++++++++++++++++++++++++++
xen/arch/x86/traps.c | 8 +++-
xen/include/asm-x86/hvm/domain.h | 3 ++
xen/include/asm-x86/hvm/hvm.h | 6 +++
xen/include/asm-x86/hvm/vmware.h | 33 +++++++++++++++++
xen/include/public/arch-x86/xen.h | 2 +-
11 files changed, 142 insertions(+), 5 deletions(-)
create mode 100644 xen/arch/x86/hvm/vmware/Makefile
create mode 100644 xen/arch/x86/hvm/vmware/cpuid.c
create mode 100644 xen/include/asm-x86/hvm/vmware.h
diff --git a/tools/libxl/libxl_x86.c b/tools/libxl/libxl_x86.c
index ed2bd38..651b338 100644
--- a/tools/libxl/libxl_x86.c
+++ b/tools/libxl/libxl_x86.c
@@ -5,8 +5,8 @@ int libxl__arch_domain_prepare_config(libxl__gc *gc,
libxl_domain_config *d_config,
xc_domain_configuration_t *xc_config)
{
- /* No specific configuration right now */
-
+ /* Note: will be changed in a later patch */
+ xc_config->vmware_hwver = 0;
return 0;
}
diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index a8fe046..c2bd52a 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -554,6 +554,7 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags,
{
d->arch.hvm_domain.hap_enabled =
hvm_funcs.hap_supported && (domcr_flags & DOMCRF_hap);
+ d->arch.hvm_domain.vmware_hwver = config->vmware_hwver;
rc = create_perdomain_mapping(d, PERDOMAIN_VIRT_START, 0, NULL, NULL);
}
diff --git a/xen/arch/x86/hvm/Makefile b/xen/arch/x86/hvm/Makefile
index 69af47f..284ca75 100644
--- a/xen/arch/x86/hvm/Makefile
+++ b/xen/arch/x86/hvm/Makefile
@@ -1,5 +1,6 @@
subdir-y += svm
subdir-y += vmx
+subdir-y += vmware
obj-y += asid.o
obj-y += emulate.o
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 535d622..c002ae7 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -59,6 +59,7 @@
#include <asm/hvm/trace.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/event.h>
+#include <asm/hvm/vmware.h>
#include <asm/mtrr.h>
#include <asm/apic.h>
#include <public/sched.h>
@@ -4289,6 +4290,9 @@ void hvm_cpuid(unsigned int input, unsigned int *eax, unsigned int *ebx,
if ( cpuid_viridian_leaves(input, eax, ebx, ecx, edx) )
return;
+ if ( cpuid_vmware_leaves(input, eax, ebx, ecx, edx) )
+ return;
+
if ( cpuid_hypervisor_leaves(input, count, eax, ebx, ecx, edx) )
return;
@@ -5693,6 +5697,13 @@ static int hvm_allow_set_param(struct domain *d,
{
/* The following parameters should only be changed once. */
case HVM_PARAM_VIRIDIAN:
+ /* Disallow if vmware_hwver is in use */
+ if ( d->arch.hvm_domain.vmware_hwver )
+ {
+ rc = -EOPNOTSUPP;
+ break;
+ }
+ /* Fall through */
case HVM_PARAM_IOREQ_SERVER_PFN:
case HVM_PARAM_NR_IOREQ_SERVER_PAGES:
if ( value != 0 && a->value != value )
diff --git a/xen/arch/x86/hvm/vmware/Makefile b/xen/arch/x86/hvm/vmware/Makefile
new file mode 100644
index 0000000..3fb2e0b
--- /dev/null
+++ b/xen/arch/x86/hvm/vmware/Makefile
@@ -0,0 +1 @@
+obj-y += cpuid.o
diff --git a/xen/arch/x86/hvm/vmware/cpuid.c b/xen/arch/x86/hvm/vmware/cpuid.c
new file mode 100644
index 0000000..0dff36b
--- /dev/null
+++ b/xen/arch/x86/hvm/vmware/cpuid.c
@@ -0,0 +1,77 @@
+/*
+ * arch/x86/hvm/vmware/cpuid.c
+ *
+ * Copyright (C) 2012 Verizon Corporation
+ *
+ * This file is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License Version 2 (GPLv2)
+ * as published by the Free Software Foundation.
+ *
+ * This file 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 for more details. <http://www.gnu.org/licenses/>.
+ */
+
+#include <xen/sched.h>
+
+#include <asm/hvm/hvm.h>
+#include <asm/hvm/vmware.h>
+
+/*
+ * VMware hardware version 7 defines some of these cpuid levels,
+ * below is a brief description about those.
+ *
+ * Leaf 0x40000000, Hypervisor CPUID information
+ * # EAX: The maximum input value for hypervisor CPUID info (0x40000010).
+ * # EBX, ECX, EDX: Hypervisor vendor ID signature. E.g. "VMwareVMware"
+ *
+ * Leaf 0x40000010, Timing information.
+ * # EAX: (Virtual) TSC frequency in kHz.
+ * # EBX: (Virtual) Bus (local apic timer) frequency in kHz.
+ * # ECX, EDX: RESERVED
+ */
+
+int cpuid_vmware_leaves(uint32_t idx, uint32_t *eax, uint32_t *ebx,
+ uint32_t *ecx, uint32_t *edx)
+{
+ struct domain *d = current->domain;
+
+ if ( !is_vmware_domain(d) ||
+ d->arch.hvm_domain.vmware_hwver < 7 )
+ return 0;
+
+ switch ( idx - 0x40000000 )
+ {
+ case 0x0:
+ *eax = 0x40000010; /* Largest leaf */
+ *ebx = 0x61774d56; /* "VMwa" */
+ *ecx = 0x4d566572; /* "reVM" */
+ *edx = 0x65726177; /* "ware" */
+ break;
+
+ case 0x10:
+ /* (Virtual) TSC frequency in kHz. */
+ *eax = d->arch.tsc_khz;
+ /* (Virtual) Bus (local apic timer) frequency in kHz. */
+ *ebx = 1000000ull / APIC_BUS_CYCLE_NS;
+ *ecx = 0; /* Reserved */
+ *edx = 0; /* Reserved */
+ break;
+
+ default:
+ return 0;
+ }
+
+ return 1;
+}
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * tab-width: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
index ac62f20..129be1c 100644
--- a/xen/arch/x86/traps.c
+++ b/xen/arch/x86/traps.c
@@ -750,8 +750,12 @@ int cpuid_hypervisor_leaves( uint32_t idx, uint32_t sub_idx,
uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx)
{
struct domain *currd = current->domain;
- /* Optionally shift out of the way of Viridian architectural leaves. */
- uint32_t base = is_viridian_domain(currd) ? 0x40000100 : 0x40000000;
+ /*
+ * Optionally shift out of the way of Viridian or VMware
+ * architectural leaves.
+ */
+ uint32_t base = is_viridian_domain(currd) || is_vmware_domain(currd) ?
+ 0x40000100 : 0x40000000;
uint32_t limit, dummy;
idx -= base;
diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
index ad68fcf..ada8aaa 100644
--- a/xen/include/asm-x86/hvm/domain.h
+++ b/xen/include/asm-x86/hvm/domain.h
@@ -110,6 +110,9 @@ struct hvm_domain {
uint64_t *params;
+ /* emulated VMware Hardware Version */
+ uint64_t vmware_hwver;
+
/* Memory ranges with pinned cache attributes. */
struct list_head pinned_cacheattr_ranges;
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 57f9605..a074afe 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -356,6 +356,12 @@ static inline unsigned long hvm_get_shadow_gs_base(struct vcpu *v)
#define has_viridian_time_ref_count(d) \
(is_viridian_domain(d) && (viridian_feature_mask(d) & HVMPV_time_ref_count))
+#define vmware_feature_mask(d) \
+ ((d)->arch.hvm_domain.vmware_hwver)
+
+#define is_vmware_domain(d) \
+ (is_hvm_domain(d) && vmware_feature_mask(d))
+
void hvm_hypervisor_cpuid_leaf(uint32_t sub_idx,
uint32_t *eax, uint32_t *ebx,
uint32_t *ecx, uint32_t *edx);
diff --git a/xen/include/asm-x86/hvm/vmware.h b/xen/include/asm-x86/hvm/vmware.h
new file mode 100644
index 0000000..8390173
--- /dev/null
+++ b/xen/include/asm-x86/hvm/vmware.h
@@ -0,0 +1,33 @@
+/*
+ * asm-x86/hvm/vmware.h
+ *
+ * Copyright (C) 2012 Verizon Corporation
+ *
+ * This file is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License Version 2 (GPLv2)
+ * as published by the Free Software Foundation.
+ *
+ * This file 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 for more details. <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef ASM_X86_HVM_VMWARE_H__
+#define ASM_X86_HVM_VMWARE_H__
+
+#include <xen/types.h>
+
+int cpuid_vmware_leaves(uint32_t idx, uint32_t *eax, uint32_t *ebx,
+ uint32_t *ecx, uint32_t *edx);
+
+#endif /* ASM_X86_HVM_VMWARE_H__ */
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/include/public/arch-x86/xen.h b/xen/include/public/arch-x86/xen.h
index 2ecc9c9..f84d10d 100644
--- a/xen/include/public/arch-x86/xen.h
+++ b/xen/include/public/arch-x86/xen.h
@@ -268,7 +268,7 @@ typedef struct arch_shared_info arch_shared_info_t;
* XEN_DOMCTL_INTERFACE_VERSION.
*/
struct xen_arch_domainconfig {
- char dummy;
+ uint64_t vmware_hwver;
};
#endif
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 3/8] tools: Add vmware_hwver support
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
2015-06-27 23:27 ` [PATCH v12 1/8] tools: Add vga=vmware Don Slutz
2015-06-27 23:27 ` [PATCH v12 2/8] xen: Add support for VMware cpuid leaves Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-07-01 20:14 ` Konrad Rzeszutek Wilk
2015-06-27 23:27 ` [PATCH v12 4/8] vmware: Add VMware provided include file Don Slutz
` (5 subsequent siblings)
8 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
This is used to set xen_arch_domainconfig vmware_hw. It is set to
the emulated VMware virtual hardware version.
Currently 0, 3-4, 6-11 are good values. However the code only
checks for == 0, != 0, or < 7.
Signed-off-by: Don Slutz <dslutz@verizon.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v12:
I'm not sure this hunk has anything to do with this patch, nor
what the semantic difference between the old and new text is
supposed to be.
Dropped comment change.
v11:
Dropped "If non-zero then default VGA to VMware's VGA"
v10:
LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE &
LIBXL_HAVE_BUILDINFO_HVM_VMWARE_HWVER are arriving together
a single umbrella could be used.
Since I split the LIBXL_VGA_INTERFACE_TYPE_VMWARE into
it's own patch, this is not longer true.
But I did use 1 for the 2 c_info changes.
Please use GCSPRINTF.
Remove vga=vmware from here.
v9:
I assumed that s/vmware_hw/vmware_hwver/ is not a big enough
change to drop the Reviewed-by. Did a minor edit to the
commit message to add 7 to the list of values checked.
v7:
Default handling of hvm.vga.kind bad.
Fixed.
Default of vmware_port should be based on vmware_hw.
Done.
v5:
Anything looking for Xen according to the Xen cpuid instructions...
Adjusted doc to new wording.
docs/man/xl.cfg.pod.5 | 17 +++++++++++++++++
tools/libxl/libxl_create.c | 4 +++-
tools/libxl/libxl_types.idl | 1 +
tools/libxl/libxl_x86.c | 3 +--
tools/libxl/xl_cmdimpl.c | 2 ++
5 files changed, 24 insertions(+), 3 deletions(-)
diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
index 84078f6..4a01527 100644
--- a/docs/man/xl.cfg.pod.5
+++ b/docs/man/xl.cfg.pod.5
@@ -1348,6 +1348,23 @@ The viridian option can be specified as a boolean. A value of true (1)
is equivalent to the list [ "defaults" ], and a value of false (0) is
equivalent to an empty list.
+=item B<vmware_hwver=NUMBER>
+
+Turns on or off the exposure of VMware cpuid. The number is
+VMware's hardware version number, where 0 is off. A number >= 7
+is needed to enable exposure of VMware cpuid.
+
+The hardware version number (vmware_hwver) comes from VMware config files.
+
+=over 4
+
+In a .vmx it is virtualHW.version
+
+In a .ovf it is part of the value of vssd:VirtualSystemType.
+For vssd:VirtualSystemType == vmx-07, vmware_hwver = 7.
+
+=back
+
=back
=head3 Emulated VGA Graphics Device
diff --git a/tools/libxl/libxl_create.c b/tools/libxl/libxl_create.c
index f366a09..02931dd 100644
--- a/tools/libxl/libxl_create.c
+++ b/tools/libxl/libxl_create.c
@@ -464,7 +464,7 @@ int libxl__domain_build(libxl__gc *gc,
vments[4] = "start_time";
vments[5] = libxl__sprintf(gc, "%lu.%02d", start_time.tv_sec,(int)start_time.tv_usec/10000);
- localents = libxl__calloc(gc, 9, sizeof(char *));
+ localents = libxl__calloc(gc, 11, sizeof(char *));
i = 0;
localents[i++] = "platform/acpi";
localents[i++] = libxl_defbool_val(info->u.hvm.acpi) ? "1" : "0";
@@ -472,6 +472,8 @@ int libxl__domain_build(libxl__gc *gc,
localents[i++] = libxl_defbool_val(info->u.hvm.acpi_s3) ? "1" : "0";
localents[i++] = "platform/acpi_s4";
localents[i++] = libxl_defbool_val(info->u.hvm.acpi_s4) ? "1" : "0";
+ localents[i++] = "platform/vmware_hwver";
+ localents[i++] = GCSPRINTF("%"PRId64, d_config->c_info.vmware_hwver);
if (info->u.hvm.mmio_hole_memkb) {
uint64_t max_ram_below_4g =
(1ULL << 32) - (info->u.hvm.mmio_hole_memkb << 10);
diff --git a/tools/libxl/libxl_types.idl b/tools/libxl/libxl_types.idl
index 534e7f0..6c0a1c7 100644
--- a/tools/libxl/libxl_types.idl
+++ b/tools/libxl/libxl_types.idl
@@ -346,6 +346,7 @@ libxl_domain_create_info = Struct("domain_create_info",[
("run_hotplug_scripts",libxl_defbool),
("pvh", libxl_defbool),
("driver_domain",libxl_defbool),
+ ("vmware_hwver", uint64),
], dir=DIR_IN)
libxl_domain_restore_params = Struct("domain_restore_params", [
diff --git a/tools/libxl/libxl_x86.c b/tools/libxl/libxl_x86.c
index 651b338..fd7dafa 100644
--- a/tools/libxl/libxl_x86.c
+++ b/tools/libxl/libxl_x86.c
@@ -5,8 +5,7 @@ int libxl__arch_domain_prepare_config(libxl__gc *gc,
libxl_domain_config *d_config,
xc_domain_configuration_t *xc_config)
{
- /* Note: will be changed in a later patch */
- xc_config->vmware_hwver = 0;
+ xc_config->vmware_hwver = d_config->c_info.vmware_hwver;
return 0;
}
diff --git a/tools/libxl/xl_cmdimpl.c b/tools/libxl/xl_cmdimpl.c
index efb3651..138adcc 100644
--- a/tools/libxl/xl_cmdimpl.c
+++ b/tools/libxl/xl_cmdimpl.c
@@ -1387,6 +1387,8 @@ static void parse_config_data(const char *config_source,
b_info->cmdline = parse_cmdline(config);
xlu_cfg_get_defbool(config, "driver_domain", &c_info->driver_domain, 0);
+ if (!xlu_cfg_get_long(config, "vmware_hwver", &l, 1))
+ c_info->vmware_hwver = l;
switch(b_info->type) {
case LIBXL_DOMAIN_TYPE_HVM:
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 4/8] vmware: Add VMware provided include file.
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
` (2 preceding siblings ...)
2015-06-27 23:27 ` [PATCH v12 3/8] tools: Add vmware_hwver support Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-06-27 23:27 ` [PATCH v12 5/8] xen: Add vmware_port support Don Slutz
` (4 subsequent siblings)
8 siblings, 0 replies; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
This file: backdoor_def.h comes from:
http://packages.vmware.com/tools/esx/3.5latest/rhel4/SRPMS/index.html
open-vm-tools-kmod-7.4.8-396269.423167.src.rpm
open-vm-tools-kmod-7.4.8.tar.gz
vmhgfs/backdoor_def.h
and is unchanged.
Added the badly named include file includeCheck.h also. It only has
a comment and is provided so that backdoor_def.h can be used without
change.
Signed-off-by: Don Slutz <dslutz@verizon.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v11,v12:
No change
v10:
Add Acked-by: Andrew Cooper
v9:
Either the description is wrong, or the patch is stale.
stale commit message -- fixed.
I'd say a file with a single comment line in it would suffice.
Done.
xen/arch/x86/hvm/vmware/backdoor_def.h | 167 +++++++++++++++++++++++++++++++++
xen/arch/x86/hvm/vmware/includeCheck.h | 1 +
2 files changed, 168 insertions(+)
create mode 100644 xen/arch/x86/hvm/vmware/backdoor_def.h
create mode 100644 xen/arch/x86/hvm/vmware/includeCheck.h
diff --git a/xen/arch/x86/hvm/vmware/backdoor_def.h b/xen/arch/x86/hvm/vmware/backdoor_def.h
new file mode 100644
index 0000000..e76795f
--- /dev/null
+++ b/xen/arch/x86/hvm/vmware/backdoor_def.h
@@ -0,0 +1,167 @@
+/* **********************************************************
+ * Copyright 1998 VMware, Inc. All rights reserved.
+ * **********************************************************
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation version 2 and no later version.
+ *
+ * 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
+ * for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/*
+ * backdoor_def.h --
+ *
+ * This contains backdoor defines that can be included from
+ * an assembly language file.
+ */
+
+
+
+#ifndef _BACKDOOR_DEF_H_
+#define _BACKDOOR_DEF_H_
+
+#define INCLUDE_ALLOW_MODULE
+#define INCLUDE_ALLOW_USERLEVEL
+#define INCLUDE_ALLOW_VMMEXT
+#define INCLUDE_ALLOW_VMCORE
+#define INCLUDE_ALLOW_VMKERNEL
+#include "includeCheck.h"
+
+/*
+ * If you want to add a new low-level backdoor call for a guest userland
+ * application, please consider using the GuestRpc mechanism instead. --hpreg
+ */
+
+#define BDOOR_MAGIC 0x564D5868
+
+/* Low-bandwidth backdoor port. --hpreg */
+
+#define BDOOR_PORT 0x5658
+
+#define BDOOR_CMD_GETMHZ 1
+/*
+ * BDOOR_CMD_APMFUNCTION is used by:
+ *
+ * o The FrobOS code, which instead should either program the virtual chipset
+ * (like the new BIOS code does, matthias offered to implement that), or not
+ * use any VM-specific code (which requires that we correctly implement
+ * "power off on CLI HLT" for SMP VMs, boris offered to implement that)
+ *
+ * o The old BIOS code, which will soon be jettisoned
+ *
+ * --hpreg
+ */
+#define BDOOR_CMD_APMFUNCTION 2
+#define BDOOR_CMD_GETDISKGEO 3
+#define BDOOR_CMD_GETPTRLOCATION 4
+#define BDOOR_CMD_SETPTRLOCATION 5
+#define BDOOR_CMD_GETSELLENGTH 6
+#define BDOOR_CMD_GETNEXTPIECE 7
+#define BDOOR_CMD_SETSELLENGTH 8
+#define BDOOR_CMD_SETNEXTPIECE 9
+#define BDOOR_CMD_GETVERSION 10
+#define BDOOR_CMD_GETDEVICELISTELEMENT 11
+#define BDOOR_CMD_TOGGLEDEVICE 12
+#define BDOOR_CMD_GETGUIOPTIONS 13
+#define BDOOR_CMD_SETGUIOPTIONS 14
+#define BDOOR_CMD_GETSCREENSIZE 15
+#define BDOOR_CMD_MONITOR_CONTROL 16
+#define BDOOR_CMD_GETHWVERSION 17
+#define BDOOR_CMD_OSNOTFOUND 18
+#define BDOOR_CMD_GETUUID 19
+#define BDOOR_CMD_GETMEMSIZE 20
+#define BDOOR_CMD_HOSTCOPY 21 /* Devel only */
+/* BDOOR_CMD_GETOS2INTCURSOR, 22, is very old and defunct. Reuse. */
+#define BDOOR_CMD_GETTIME 23 /* Deprecated. Use GETTIMEFULL. */
+#define BDOOR_CMD_STOPCATCHUP 24
+#define BDOOR_CMD_PUTCHR 25 /* Devel only */
+#define BDOOR_CMD_ENABLE_MSG 26 /* Devel only */
+#define BDOOR_CMD_GOTO_TCL 27 /* Devel only */
+#define BDOOR_CMD_INITPCIOPROM 28
+#define BDOOR_CMD_INT13 29
+#define BDOOR_CMD_MESSAGE 30
+#define BDOOR_CMD_RSVD0 31
+#define BDOOR_CMD_RSVD1 32
+#define BDOOR_CMD_RSVD2 33
+#define BDOOR_CMD_ISACPIDISABLED 34
+#define BDOOR_CMD_TOE 35 /* Not in use */
+/* BDOOR_CMD_INITLSIOPROM, 36, was merged with 28. Reuse. */
+#define BDOOR_CMD_PATCH_SMBIOS_STRUCTS 37
+#define BDOOR_CMD_MAPMEM 38 /* Devel only */
+#define BDOOR_CMD_ABSPOINTER_DATA 39
+#define BDOOR_CMD_ABSPOINTER_STATUS 40
+#define BDOOR_CMD_ABSPOINTER_COMMAND 41
+#define BDOOR_CMD_TIMER_SPONGE 42
+#define BDOOR_CMD_PATCH_ACPI_TABLES 43
+/* Catch-all to allow synchronous tests */
+#define BDOOR_CMD_DEVEL_FAKEHARDWARE 44 /* Debug only - needed in beta */
+#define BDOOR_CMD_GETHZ 45
+#define BDOOR_CMD_GETTIMEFULL 46
+#define BDOOR_CMD_STATELOGGER 47
+#define BDOOR_CMD_CHECKFORCEBIOSSETUP 48
+#define BDOOR_CMD_LAZYTIMEREMULATION 49
+#define BDOOR_CMD_BIOSBBS 50
+#define BDOOR_CMD_MAX 51
+
+/*
+ * IMPORTANT NOTE: When modifying the behavior of an existing backdoor command,
+ * you must adhere to the semantics expected by the oldest Tools who use that
+ * command. Specifically, do not alter the way in which the command modifies
+ * the registers. Otherwise backwards compatibility will suffer.
+ */
+
+/* High-bandwidth backdoor port. --hpreg */
+
+#define BDOORHB_PORT 0x5659
+
+#define BDOORHB_CMD_MESSAGE 0
+#define BDOORHB_CMD_MAX 1
+
+/*
+ * There is another backdoor which allows access to certain TSC-related
+ * values using otherwise illegal PMC indices when the pseudo_perfctr
+ * control flag is set.
+ */
+
+#define BDOOR_PMC_HW_TSC 0x10000
+#define BDOOR_PMC_REAL_NS 0x10001
+#define BDOOR_PMC_APPARENT_NS 0x10002
+
+#define IS_BDOOR_PMC(index) (((index) | 3) == 0x10003)
+#define BDOOR_CMD(ecx) ((ecx) & 0xffff)
+
+
+#ifdef VMM
+/*
+ *----------------------------------------------------------------------
+ *
+ * Backdoor_CmdRequiresFullyValidVCPU --
+ *
+ * A few backdoor commands require the full VCPU to be valid
+ * (including GDTR, IDTR, TR and LDTR). The rest get read/write
+ * access to GPRs and read access to Segment registers (selectors).
+ *
+ * Result:
+ * True iff VECX contains a command that require the full VCPU to
+ * be valid.
+ *
+ *----------------------------------------------------------------------
+ */
+static INLINE Bool
+Backdoor_CmdRequiresFullyValidVCPU(unsigned cmd)
+{
+ return cmd == BDOOR_CMD_RSVD0 ||
+ cmd == BDOOR_CMD_RSVD1 ||
+ cmd == BDOOR_CMD_RSVD2;
+}
+#endif
+
+#endif
diff --git a/xen/arch/x86/hvm/vmware/includeCheck.h b/xen/arch/x86/hvm/vmware/includeCheck.h
new file mode 100644
index 0000000..3b63fa4
--- /dev/null
+++ b/xen/arch/x86/hvm/vmware/includeCheck.h
@@ -0,0 +1 @@
+/* Nothing here. Just to use backdoor_def.h without change. */
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 5/8] xen: Add vmware_port support
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
` (3 preceding siblings ...)
2015-06-27 23:27 ` [PATCH v12 4/8] vmware: Add VMware provided include file Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-06-27 23:27 ` [PATCH v12 6/8] tools: " Don Slutz
` (3 subsequent siblings)
8 siblings, 0 replies; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
This includes adding is_vmware_port_enabled
This is a new xen_arch_domainconfig flag,
XEN_DOMCTL_CONFIG_VMWARE_PORT_MASK.
This enables limited support of VMware's hyper-call.
This is both a more complete support then in currently provided by
QEMU and/or KVM and less. The missing part requires QEMU changes
and has been left out until the QEMU patches are accepted upstream.
VMware's hyper-call is also known as VMware Backdoor I/O Port.
Note: this support does not depend on vmware_hw being non-zero.
Summary is that VMware treats "in (%dx),%eax" (or "out %eax,(%dx)")
to port 0x5658 specially. Note: since many operations return data
in EAX, "in (%dx),%eax" is the one to use. The other lengths like
"in (%dx),%al" will still do things, only AL part of EAX will be
changed. For "out %eax,(%dx)" of all lengths, EAX will remain
unchanged.
An open source example of using this is:
http://open-vm-tools.sourceforge.net/
Which only uses "inl (%dx)". Also
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1009458
Some of the best info is at:
https://sites.google.com/site/chitchatvmback/backdoor
Signed-off-by: Don Slutz <dslutz@verizon.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v12:
Surrounding code avoiding the use of "break" makes the result
look rather inconsistent. Please move this up immediately after
the XSM check, or drop the "break".
Moved it up.
v11:
Dropped ASSERT(is_hvm_domain(currd))
Newline after break;
v10:
Probably better as EOPNOTSUPP, as it is a configuration problem.
This function looks as if it should be static.
I would suggest putting vmport_register declaration in hvm.h ...
As indicated before, I don't think this is a good use case for a
domain creation flag.
Switch to the new config way.
struct domain *d => struct domain *currd
Are you sure you don't want to zero the high halves of 64-bit ...
Comment added.
Then just have this handled into the default case.
Reworked new_eax handling.
is_hvm_domain(currd)
And - why here rather than before the switch() or even right at the
start of the function?
Moved to start.
With that, is it really correct that OUT updates the other registers
just like IN? If so, this deserves a comment, so that readers won't
think this is in error.
All done in comment at start.
v9:
Switch to x86_emulator to handle #GP code moved to next patch.
Can you explain why a HVM param isn't suitable here?
Issue with changing QEMU on the fly.
Andrew Cooper: My recommendation is still to use a creation flag
So no change.
Please move SVM's identical definition into ...
Did this as #1. No longer needed, but since the patch was ready
I have included it.
--Lots of questions about code that no long is part of this patch. --
With this, is handling other than 32-bit in/out really
meaningful/correct?
Added comment about this.
Since you can't get here for PV, I can't see what you need this.
Changed to an ASSERT.
Why version 4?
Added comment about this.
-- Several questions about register changes.
Re-coded to use new_eax and set *val to this.
Change to generealy use reg->_e..
These ei1/ei2 checks belong in the callers imo -
Moved.
the "port" function parameter isn't even checked
Add check for exact match.
If dropping the code is safe without also forbidding the
combination of nested and VMware emulation.
Added the forbidding the combination of nested and VMware.
Mostly do to the cases of the nested virtual code is the one
to handle VMware stuff if needed, not the root one. Also I am
having issues testing xen nested in xen and using hvm.
v7:
More on AMD in the commit message.
Switch to only change 32bit part of registers, what VMware
does.
Too much logging and tracing.
Dropped a lot of it. This includes vmport_debug=
v6:
Dropped the attempt to use svm_nextrip_insn_length via
__get_instruction_length (added in v2). Just always look
at upto 15 bytes on AMD.
v5:
we should make sure that svm_vmexit_gp_intercept is not executed for
any other guest.
Added an ASSERT on is_vmware_port_enabled.
magic integers?
Added #define for them.
I am fairly certain that you need some brackets here.
Added brackets.
xen/arch/x86/domain.c | 2 +
xen/arch/x86/hvm/hvm.c | 9 +++
xen/arch/x86/hvm/vmware/Makefile | 1 +
xen/arch/x86/hvm/vmware/vmport.c | 148 ++++++++++++++++++++++++++++++++++++++
xen/include/asm-x86/hvm/domain.h | 3 +
xen/include/asm-x86/hvm/hvm.h | 2 +
xen/include/public/arch-x86/xen.h | 6 ++
7 files changed, 171 insertions(+)
create mode 100644 xen/arch/x86/hvm/vmware/vmport.c
diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index c2bd52a..39e323b 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -555,6 +555,8 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags,
d->arch.hvm_domain.hap_enabled =
hvm_funcs.hap_supported && (domcr_flags & DOMCRF_hap);
d->arch.hvm_domain.vmware_hwver = config->vmware_hwver;
+ d->arch.hvm_domain.is_vmware_port_enabled =
+ !!(config->arch_flags & XEN_DOMCTL_CONFIG_VMWARE_PORT_MASK);
rc = create_perdomain_mapping(d, PERDOMAIN_VIRT_START, 0, NULL, NULL);
}
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index c002ae7..f8ec170 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -1486,6 +1486,9 @@ int hvm_domain_initialise(struct domain *d)
else
d->arch.hvm_domain.io_bitmap = hvm_io_bitmap;
+ if ( d->arch.hvm_domain.is_vmware_port_enabled )
+ vmport_register(d);
+
if ( is_pvh_domain(d) )
{
register_portio_handler(d, 0, 0x10003, handle_pvh_io);
@@ -5820,6 +5823,12 @@ static int hvmop_set_param(
rc = xsm_hvm_param_nested(XSM_PRIV, d);
if ( rc )
break;
+ /* Prevent nestedhvm with vmport */
+ if ( d->arch.hvm_domain.is_vmware_port_enabled )
+ {
+ rc = -EOPNOTSUPP;
+ break;
+ }
if ( a.value > 1 )
rc = -EINVAL;
/*
diff --git a/xen/arch/x86/hvm/vmware/Makefile b/xen/arch/x86/hvm/vmware/Makefile
index 3fb2e0b..cd8815b 100644
--- a/xen/arch/x86/hvm/vmware/Makefile
+++ b/xen/arch/x86/hvm/vmware/Makefile
@@ -1 +1,2 @@
obj-y += cpuid.o
+obj-y += vmport.o
diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
new file mode 100644
index 0000000..f24d8e3
--- /dev/null
+++ b/xen/arch/x86/hvm/vmware/vmport.c
@@ -0,0 +1,148 @@
+/*
+ * HVM VMPORT emulation
+ *
+ * Copyright (C) 2012 Verizon Corporation
+ *
+ * This file is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License Version 2 (GPLv2)
+ * as published by the Free Software Foundation.
+ *
+ * This file 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 for more details. <http://www.gnu.org/licenses/>.
+ */
+
+#include <xen/lib.h>
+#include <asm/hvm/hvm.h>
+#include <asm/hvm/support.h>
+
+#include "backdoor_def.h"
+
+static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
+{
+ struct cpu_user_regs *regs = guest_cpu_user_regs();
+
+ /*
+ * While VMware expects only 32-bit in, they do support using
+ * other sizes and out. However they do require only the 1 port
+ * and the correct value in eax. Since some of the data
+ * returned in eax is smaller the 32 bits and/or you only need
+ * the other registers the dir and bytes do not need any
+ * checking. The caller will handle the bytes, and dir is
+ * handled below for eax.
+ */
+ if ( port == BDOOR_PORT && regs->_eax == BDOOR_MAGIC )
+ {
+ uint32_t new_eax = ~0u;
+ uint64_t value;
+ struct vcpu *curr = current;
+ struct domain *currd = curr->domain;
+
+ /*
+ * VMware changes the other (non eax) registers ignoring dir
+ * (IN vs OUT). It also changes only the 32-bit part
+ * leaving the high 32-bits unchanged, unlike what one would
+ * expect to happen.
+ */
+ switch ( regs->_ecx & 0xffff )
+ {
+ case BDOOR_CMD_GETMHZ:
+ new_eax = currd->arch.tsc_khz / 1000;
+ break;
+
+ case BDOOR_CMD_GETVERSION:
+ /* MAGIC */
+ regs->_ebx = BDOOR_MAGIC;
+ /* VERSION_MAGIC */
+ new_eax = 6;
+ /* Claim we are an ESX. VMX_TYPE_SCALABLE_SERVER */
+ regs->_ecx = 2;
+ break;
+
+ case BDOOR_CMD_GETHWVERSION:
+ /* vmware_hw */
+ new_eax = currd->arch.hvm_domain.vmware_hwver;
+ /*
+ * Returning zero is not the best. VMware was not at
+ * all consistent in the handling of this command until
+ * VMware hardware version 4. So it is better to claim
+ * 4 then 0. This should only happen in strange configs.
+ */
+ if ( !new_eax )
+ new_eax = 4;
+ break;
+
+ case BDOOR_CMD_GETHZ:
+ {
+ struct segment_register sreg;
+
+ hvm_get_segment_register(curr, x86_seg_ss, &sreg);
+ if ( sreg.attr.fields.dpl == 0 )
+ {
+ value = currd->arch.tsc_khz * 1000;
+ /* apic-frequency (bus speed) */
+ regs->_ecx = 1000000000ULL / APIC_BUS_CYCLE_NS;
+ /* High part of tsc-frequency */
+ regs->_ebx = value >> 32;
+ /* Low part of tsc-frequency */
+ new_eax = value;
+ }
+ break;
+
+ }
+ case BDOOR_CMD_GETTIME:
+ value = get_localtime_us(currd) -
+ currd->time_offset_seconds * 1000000ULL;
+ /* hostUsecs */
+ regs->_ebx = value % 1000000UL;
+ /* hostSecs */
+ new_eax = value / 1000000ULL;
+ /* maxTimeLag */
+ regs->_ecx = 1000000;
+ /* offset to GMT in minutes */
+ regs->_edx = currd->time_offset_seconds / 60;
+ break;
+
+ case BDOOR_CMD_GETTIMEFULL:
+ /* BDOOR_MAGIC */
+ new_eax = BDOOR_MAGIC;
+ value = get_localtime_us(currd) -
+ currd->time_offset_seconds * 1000000ULL;
+ /* hostUsecs */
+ regs->_ebx = value % 1000000UL;
+ /* hostSecs low 32 bits */
+ regs->_edx = value / 1000000ULL;
+ /* hostSecs high 32 bits */
+ regs->_esi = (value / 1000000ULL) >> 32;
+ /* maxTimeLag */
+ regs->_ecx = 1000000;
+ break;
+
+ default:
+ /* Let backing DM handle */
+ return X86EMUL_UNHANDLEABLE;
+ }
+ if ( dir == IOREQ_READ )
+ *val = new_eax;
+ }
+ else if ( dir == IOREQ_READ )
+ *val = ~0u;
+
+ return X86EMUL_OKAY;
+}
+
+void vmport_register(struct domain *d)
+{
+ register_portio_handler(d, BDOOR_PORT, 4, vmport_ioport);
+}
+
+/*
+ * Local variables:
+ * mode: C
+ * c-set-style: "BSD"
+ * c-basic-offset: 4
+ * tab-width: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
index ada8aaa..726b977 100644
--- a/xen/include/asm-x86/hvm/domain.h
+++ b/xen/include/asm-x86/hvm/domain.h
@@ -125,6 +125,9 @@ struct hvm_domain {
spinlock_t uc_lock;
bool_t is_in_uc_mode;
+ /* VMware backdoor port available */
+ bool_t is_vmware_port_enabled;
+
/* Pass-through */
struct hvm_iommu hvm_iommu;
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index a074afe..394c230 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -522,6 +522,8 @@ extern bool_t opt_hvm_fep;
#define opt_hvm_fep 0
#endif
+void vmport_register(struct domain *d);
+
#endif /* __ASM_X86_HVM_HVM_H__ */
/*
diff --git a/xen/include/public/arch-x86/xen.h b/xen/include/public/arch-x86/xen.h
index f84d10d..53b84da 100644
--- a/xen/include/public/arch-x86/xen.h
+++ b/xen/include/public/arch-x86/xen.h
@@ -267,8 +267,14 @@ typedef struct arch_shared_info arch_shared_info_t;
* struct xen_arch_domainconfig's ABI is covered by
* XEN_DOMCTL_INTERFACE_VERSION.
*/
+
+/* Enable use of vmware backdoor port. */
+#define XEN_DOMCTL_CONFIG_VMWARE_PORT_BIT 0
+#define XEN_DOMCTL_CONFIG_VMWARE_PORT_MASK (1U << XEN_DOMCTL_CONFIG_VMWARE_PORT_BIT)
+
struct xen_arch_domainconfig {
uint64_t vmware_hwver;
+ uint64_t arch_flags;
};
#endif
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 6/8] tools: Add vmware_port support
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
` (4 preceding siblings ...)
2015-06-27 23:27 ` [PATCH v12 5/8] xen: Add vmware_port support Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-07-07 16:01 ` Ian Campbell
2015-06-27 23:27 ` [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT Don Slutz
` (2 subsequent siblings)
8 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
This new libxl_domain_create_info field is used to set
XEN_DOMCTL_CONFIG_VMWARE_PORT_MASK in the xc_domain_configuration_t
for x86.
In xen it is is_vmware_port_enabled.
If is_vmware_port_enabled then
enable a limited support of VMware's hyper-call.
VMware's hyper-call is also known as VMware Backdoor I/O Port.
if vmware_port is not specified in the config file, let
"vmware_hwver != 0" be the default value. This means that only
vmware_hwver = 7 needs to be specified to enable both features.
vmware_hwver = 7 is special because that is what controls the
enable of CPUID leaves for VMware (vmware_hwver >= 7).
Note: vmware_port and nestedhvm cannot be specified at the
same time.
Signed-off-by: Don Slutz <dslutz@verizon.com>
---
v12:
s/come/comes/
In v11 this seems to have morphed into only
LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE being provided, which
is clearly not an appropriate umbrella #define.
"#define LIBXL_HAVE_CREATEINFO_VMWARE 1"
Lets just have a single one of these indicating support for
vmware, it should be added at the end of the series after all
the baseline vmware functionality is in place. I think that
means hwver, vga=vmware and this port stuff.
Make (tools: Add vga=vmware) no longer independent.
Change the #define to "LIBXL_HAVE_VMWARE"
v11:
Dropped "If non-zero then default VGA to VMware's VGA"
v10:
If..." at the start of the sentence ...
Also, why is 7 special?
docs/man/xl.cfg.pod.5 | 15 +++++++++++++++
tools/libxl/libxl.h | 5 +++++
tools/libxl/libxl_create.c | 9 +++++++++
tools/libxl/libxl_types.idl | 1 +
tools/libxl/libxl_x86.c | 2 ++
tools/libxl/xl_cmdimpl.c | 1 +
6 files changed, 33 insertions(+)
diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
index 4a01527..8fb5d72 100644
--- a/docs/man/xl.cfg.pod.5
+++ b/docs/man/xl.cfg.pod.5
@@ -1354,6 +1354,8 @@ Turns on or off the exposure of VMware cpuid. The number is
VMware's hardware version number, where 0 is off. A number >= 7
is needed to enable exposure of VMware cpuid.
+If not zero it changes the default for vmware_port to on.
+
The hardware version number (vmware_hwver) comes from VMware config files.
=over 4
@@ -1365,6 +1367,19 @@ For vssd:VirtualSystemType == vmx-07, vmware_hwver = 7.
=back
+=item B<vmware_port=BOOLEAN>
+
+Turns on or off the exposure of VMware port. This is known as
+vmport in QEMU. Also called VMware Backdoor I/O Port. Not all
+defined VMware backdoor commands are implemented. All of the
+ones that Linux kernel uses are defined.
+
+Defaults to enabled if vmware_hwver is non-zero (i.e. enabled)
+otherwise defaults to disabled.
+
+Note: vmware_port and nestedhvm cannot be specified at the
+same time.
+
=back
=head3 Emulated VGA Graphics Device
diff --git a/tools/libxl/libxl.h b/tools/libxl/libxl.h
index a1c5d15..d33a509 100644
--- a/tools/libxl/libxl.h
+++ b/tools/libxl/libxl.h
@@ -200,6 +200,11 @@
#define LIBXL_HAVE_DEVICETREE_PASSTHROUGH 1
/*
+ * libxl has VMware changes.
+ */
+#define LIBXL_HAVE_VMWARE 1
+
+/*
* libxl ABI compatibility
*
* The only guarantee which libxl makes regarding ABI compatibility
diff --git a/tools/libxl/libxl_create.c b/tools/libxl/libxl_create.c
index 02931dd..eff145b 100644
--- a/tools/libxl/libxl_create.c
+++ b/tools/libxl/libxl_create.c
@@ -41,6 +41,7 @@ int libxl__domain_create_info_setdefault(libxl__gc *gc,
libxl_defbool_setdefault(&c_info->hap, libxl_defbool_val(c_info->pvh));
}
+ libxl_defbool_setdefault(&c_info->vmware_port, c_info->vmware_hwver != 0);
libxl_defbool_setdefault(&c_info->run_hotplug_scripts, true);
libxl_defbool_setdefault(&c_info->driver_domain, false);
@@ -917,6 +918,14 @@ static void initiate_domain_create(libxl__egc *egc,
ret = libxl__domain_build_info_setdefault(gc, &d_config->b_info);
if (ret) goto error_out;
+ if (d_config->c_info.type == LIBXL_DOMAIN_TYPE_HVM &&
+ libxl_defbool_val(d_config->b_info.u.hvm.nested_hvm) &&
+ libxl_defbool_val(d_config->c_info.vmware_port)) {
+ LOG(ERROR,
+ "vmware_port and nestedhvm cannot be enabled simultaneously\n");
+ ret = ERROR_INVAL;
+ goto error_out;
+ }
if (!sched_params_valid(gc, domid, &d_config->b_info.sched_params)) {
LOG(ERROR, "Invalid scheduling parameters\n");
ret = ERROR_INVAL;
diff --git a/tools/libxl/libxl_types.idl b/tools/libxl/libxl_types.idl
index 6c0a1c7..78554be 100644
--- a/tools/libxl/libxl_types.idl
+++ b/tools/libxl/libxl_types.idl
@@ -347,6 +347,7 @@ libxl_domain_create_info = Struct("domain_create_info",[
("pvh", libxl_defbool),
("driver_domain",libxl_defbool),
("vmware_hwver", uint64),
+ ("vmware_port", libxl_defbool),
], dir=DIR_IN)
libxl_domain_restore_params = Struct("domain_restore_params", [
diff --git a/tools/libxl/libxl_x86.c b/tools/libxl/libxl_x86.c
index fd7dafa..404904a 100644
--- a/tools/libxl/libxl_x86.c
+++ b/tools/libxl/libxl_x86.c
@@ -6,6 +6,8 @@ int libxl__arch_domain_prepare_config(libxl__gc *gc,
xc_domain_configuration_t *xc_config)
{
xc_config->vmware_hwver = d_config->c_info.vmware_hwver;
+ if (libxl_defbool_val(d_config->c_info.vmware_port))
+ xc_config->arch_flags |= XEN_DOMCTL_CONFIG_VMWARE_PORT_MASK;
return 0;
}
diff --git a/tools/libxl/xl_cmdimpl.c b/tools/libxl/xl_cmdimpl.c
index 138adcc..c48d52a 100644
--- a/tools/libxl/xl_cmdimpl.c
+++ b/tools/libxl/xl_cmdimpl.c
@@ -1234,6 +1234,7 @@ static void parse_config_data(const char *config_source,
}
xlu_cfg_get_defbool(config, "oos", &c_info->oos, 0);
+ xlu_cfg_get_defbool(config, "vmware_port", &c_info->vmware_port, 0);
if (!xlu_cfg_get_string (config, "pool", &buf, 0))
xlu_cfg_replace_string(config, "pool", &c_info->pool_name, 0);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
` (5 preceding siblings ...)
2015-06-27 23:27 ` [PATCH v12 6/8] tools: " Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-07-01 20:49 ` Konrad Rzeszutek Wilk
2015-06-27 23:27 ` [PATCH v12 8/8] Add xentrace to vmware_port Don Slutz
2015-07-10 8:40 ` [PATCH v12 0/8] Xen VMware tools support Jan Beulich
8 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
This adds synchronization of the 6 vcpu registers (only 32bits of
them) that vmport.c needs between Xen and QEMU.
This is to avoid a 2nd and 3rd exchange between QEMU and Xen to
fetch and put these 6 vcpu registers used by the code in vmport.c
and vmmouse.c
In the tools, enable usage of QEMU's vmport code.
The currently most useful VMware port support that QEMU has is the
VMware mouse support. Xorg included a VMware mouse support that
uses absolute mode. This make using a mouse in X11 much nicer.
Signed-off-by: Don Slutz <dslutz@verizon.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v12:
Rebase changes.
Pass size to vmport_check_port() -- required if overlap
I.E. inl on port 0x5657, 0x5656, 0x5655, 0x5659, 0x565a,
and 0x565b.
Move define of vmport_check_port() into this patch from ring3
patch.
v11:
No change
v10:
These literals should become an enum.
I don't think the invalidate type is needed.
Code handling "case X86EMUL_UNHANDLEABLE:" in emulate.c
is unclear.
Comment about "special' range of 1" is not clear.
v9:
New code was presented as an RFC before this.
Paul Durrant sugested I add support for other IOREQ types
to HVMOP_map_io_range_to_ioreq_server.
I have done this.
tools/libxc/xc_hvm_build_x86.c | 5 +-
tools/libxl/libxl_dm.c | 2 +
xen/arch/x86/hvm/emulate.c | 83 +++++++++++++++---
xen/arch/x86/hvm/hvm.c | 182 ++++++++++++++++++++++++++++++++++-----
xen/arch/x86/hvm/io.c | 16 ++++
xen/arch/x86/hvm/vmware/vmport.c | 13 +++
xen/include/asm-x86/hvm/domain.h | 3 +-
xen/include/asm-x86/hvm/hvm.h | 2 +
xen/include/public/hvm/hvm_op.h | 5 ++
xen/include/public/hvm/ioreq.h | 17 ++++
xen/include/public/hvm/params.h | 4 +-
11 files changed, 293 insertions(+), 39 deletions(-)
diff --git a/tools/libxc/xc_hvm_build_x86.c b/tools/libxc/xc_hvm_build_x86.c
index 003ea06..feebbf7 100644
--- a/tools/libxc/xc_hvm_build_x86.c
+++ b/tools/libxc/xc_hvm_build_x86.c
@@ -46,7 +46,8 @@
#define SPECIALPAGE_IOREQ 5
#define SPECIALPAGE_IDENT_PT 6
#define SPECIALPAGE_CONSOLE 7
-#define NR_SPECIAL_PAGES 8
+#define SPECIALPAGE_VMPORT_REGS 8
+#define NR_SPECIAL_PAGES 9
#define special_pfn(x) (0xff000u - NR_SPECIAL_PAGES + (x))
#define NR_IOREQ_SERVER_PAGES 8
@@ -610,6 +611,8 @@ static int setup_guest(xc_interface *xch,
special_pfn(SPECIALPAGE_BUFIOREQ));
xc_hvm_param_set(xch, dom, HVM_PARAM_IOREQ_PFN,
special_pfn(SPECIALPAGE_IOREQ));
+ xc_hvm_param_set(xch, dom, HVM_PARAM_VMPORT_REGS_PFN,
+ special_pfn(SPECIALPAGE_VMPORT_REGS));
xc_hvm_param_set(xch, dom, HVM_PARAM_CONSOLE_PFN,
special_pfn(SPECIALPAGE_CONSOLE));
xc_hvm_param_set(xch, dom, HVM_PARAM_PAGING_RING_PFN,
diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
index 975996d..966ca46 100644
--- a/tools/libxl/libxl_dm.c
+++ b/tools/libxl/libxl_dm.c
@@ -831,6 +831,8 @@ static int libxl__build_device_model_args_new(libxl__gc *gc,
machinearg, max_ram_below_4g);
}
}
+ if (libxl_defbool_val(c_info->vmware_port))
+ machinearg = GCSPRINTF("%s,vmport=on", machinearg);
flexarray_append(dm_args, machinearg);
for (i = 0; b_info->extra_hvm && b_info->extra_hvm[i] != NULL; i++)
flexarray_append(dm_args, b_info->extra_hvm[i]);
diff --git a/xen/arch/x86/hvm/emulate.c b/xen/arch/x86/hvm/emulate.c
index fe5661d..51a97d5 100644
--- a/xen/arch/x86/hvm/emulate.c
+++ b/xen/arch/x86/hvm/emulate.c
@@ -169,33 +169,88 @@ static int hvmemul_do_io(
vio->io_state = HVMIO_handle_mmio_awaiting_completion;
break;
case X86EMUL_UNHANDLEABLE:
- {
- struct hvm_ioreq_server *s =
- hvm_select_ioreq_server(curr->domain, &p);
-
- /* If there is no suitable backing DM, just ignore accesses */
- if ( !s )
+ if ( vmport_check_port(p.addr, p.size) )
{
- hvm_complete_assist_req(&p);
- rc = X86EMUL_OKAY;
- vio->io_state = HVMIO_none;
+ struct hvm_ioreq_server *s =
+ hvm_select_ioreq_server(curr->domain, &p);
+
+ /* If there is no suitable backing DM, just ignore accesses */
+ if ( !s )
+ {
+ hvm_complete_assist_req(&p);
+ rc = X86EMUL_OKAY;
+ vio->io_state = HVMIO_none;
+ }
+ else
+ {
+ rc = hvm_send_assist_req(s, &p);
+ if ( rc != X86EMUL_RETRY )
+ vio->io_state = HVMIO_none;
+ else if ( data_is_addr || dir == IOREQ_WRITE )
+ rc = X86EMUL_OKAY;
+ }
}
else
{
- rc = hvm_send_assist_req(s, &p);
- if ( rc != X86EMUL_RETRY )
- vio->io_state = HVMIO_none;
- else if ( data_is_addr || dir == IOREQ_WRITE )
+ struct hvm_ioreq_server *s;
+ vmware_regs_t *vr;
+
+ BUILD_BUG_ON(sizeof(ioreq_t) < sizeof(vmware_regs_t));
+
+ p.type = IOREQ_TYPE_VMWARE_PORT;
+ s = hvm_select_ioreq_server(curr->domain, &p);
+ vr = get_vmport_regs_any(s, curr);
+
+ /*
+ * If there is no suitable backing DM, just ignore accesses. If
+ * we do not have access to registers to pass to QEMU, just
+ * ignore access.
+ */
+ if ( !s || !vr )
+ {
+ hvm_complete_assist_req(&p);
rc = X86EMUL_OKAY;
+ vio->io_state = HVMIO_none;
+ }
+ else
+ {
+ struct cpu_user_regs *regs = guest_cpu_user_regs();
+
+ p.data = regs->rax;
+ vr->ebx = regs->_ebx;
+ vr->ecx = regs->_ecx;
+ vr->edx = regs->_edx;
+ vr->esi = regs->_esi;
+ vr->edi = regs->_edi;
+
+ vio->io_state = HVMIO_handle_pio_awaiting_completion;
+ rc = hvm_send_assist_req(s, &p);
+ if ( rc != X86EMUL_RETRY )
+ {
+ vio->io_state = HVMIO_none;
+ if ( rc == X86EMUL_OKAY )
+ rc = X86EMUL_RETRY;
+ }
+ }
}
break;
- }
default:
BUG();
}
if ( rc != X86EMUL_OKAY )
+ {
+ /*
+ * If rc is X86EMUL_RETRY and vio->io_state is
+ * HVMIO_handle_pio_awaiting_completion, then were are of
+ * type IOREQ_TYPE_VMWARE_PORT, so completion in
+ * hvm_io_assist() with no re-emulation required
+ */
+ if ( rc == X86EMUL_RETRY &&
+ vio->io_state == HVMIO_handle_pio_awaiting_completion )
+ rc = X86EMUL_OKAY;
return rc;
+ }
finish_access:
if ( dir == IOREQ_READ )
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index f8ec170..ce8cd09 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -394,6 +394,47 @@ static ioreq_t *get_ioreq(struct hvm_ioreq_server *s, struct vcpu *v)
return &p->vcpu_ioreq[v->vcpu_id];
}
+static vmware_regs_t *get_vmport_regs_one(struct hvm_ioreq_server *s,
+ struct vcpu *v)
+{
+ struct hvm_ioreq_vcpu *sv;
+
+ list_for_each_entry ( sv,
+ &s->ioreq_vcpu_list,
+ list_entry )
+ {
+ if ( sv->vcpu == v )
+ {
+ shared_vmport_iopage_t *p = s->vmport_ioreq.va;
+ if ( !p )
+ return NULL;
+ return &p->vcpu_vmport_regs[v->vcpu_id];
+ }
+ }
+ return NULL;
+}
+
+vmware_regs_t *get_vmport_regs_any(struct hvm_ioreq_server *s, struct vcpu *v)
+{
+ struct domain *d = v->domain;
+
+ ASSERT((v == current) || !vcpu_runnable(v));
+
+ if ( s )
+ return get_vmport_regs_one(s, v);
+
+ list_for_each_entry ( s,
+ &d->arch.hvm_domain.ioreq_server.list,
+ list_entry )
+ {
+ vmware_regs_t *ret = get_vmport_regs_one(s, v);
+
+ if ( ret )
+ return ret;
+ }
+ return NULL;
+}
+
bool_t hvm_io_pending(struct vcpu *v)
{
struct domain *d = v->domain;
@@ -504,22 +545,56 @@ static void hvm_free_ioreq_gmfn(struct domain *d, unsigned long gmfn)
set_bit(i, &d->arch.hvm_domain.ioreq_gmfn.mask);
}
-static void hvm_unmap_ioreq_page(struct hvm_ioreq_server *s, bool_t buf)
+typedef enum {
+ IOREQ_PAGE_TYPE_IOREQ,
+ IOREQ_PAGE_TYPE_BUFIOREQ,
+ IOREQ_PAGE_TYPE_VMPORT,
+} ioreq_page_type_t;
+
+static void hvm_unmap_ioreq_page(struct hvm_ioreq_server *s, ioreq_page_type_t buf)
{
- struct hvm_ioreq_page *iorp = buf ? &s->bufioreq : &s->ioreq;
+ struct hvm_ioreq_page *iorp = NULL;
+
+ switch ( buf )
+ {
+ case IOREQ_PAGE_TYPE_IOREQ:
+ iorp = &s->ioreq;
+ break;
+ case IOREQ_PAGE_TYPE_BUFIOREQ:
+ iorp = &s->bufioreq;
+ break;
+ case IOREQ_PAGE_TYPE_VMPORT:
+ iorp = &s->vmport_ioreq;
+ break;
+ }
+ ASSERT(iorp);
destroy_ring_for_helper(&iorp->va, iorp->page);
}
static int hvm_map_ioreq_page(
- struct hvm_ioreq_server *s, bool_t buf, unsigned long gmfn)
+ struct hvm_ioreq_server *s, ioreq_page_type_t buf, unsigned long gmfn)
{
struct domain *d = s->domain;
- struct hvm_ioreq_page *iorp = buf ? &s->bufioreq : &s->ioreq;
+ struct hvm_ioreq_page *iorp = NULL;
struct page_info *page;
void *va;
int rc;
+ switch ( buf )
+ {
+ case IOREQ_PAGE_TYPE_IOREQ:
+ iorp = &s->ioreq;
+ break;
+ case IOREQ_PAGE_TYPE_BUFIOREQ:
+ iorp = &s->bufioreq;
+ break;
+ case IOREQ_PAGE_TYPE_VMPORT:
+ iorp = &s->vmport_ioreq;
+ break;
+ }
+ ASSERT(iorp);
+
if ( (rc = prepare_ring_for_helper(d, gmfn, &page, &va)) )
return rc;
@@ -735,19 +810,32 @@ static void hvm_ioreq_server_remove_all_vcpus(struct hvm_ioreq_server *s)
static int hvm_ioreq_server_map_pages(struct hvm_ioreq_server *s,
unsigned long ioreq_pfn,
- unsigned long bufioreq_pfn)
+ unsigned long bufioreq_pfn,
+ unsigned long vmport_ioreq_pfn)
{
int rc;
- rc = hvm_map_ioreq_page(s, 0, ioreq_pfn);
+ rc = hvm_map_ioreq_page(s, IOREQ_PAGE_TYPE_IOREQ, ioreq_pfn);
if ( rc )
return rc;
if ( bufioreq_pfn != INVALID_GFN )
- rc = hvm_map_ioreq_page(s, 1, bufioreq_pfn);
+ rc = hvm_map_ioreq_page(s, IOREQ_PAGE_TYPE_BUFIOREQ, bufioreq_pfn);
if ( rc )
- hvm_unmap_ioreq_page(s, 0);
+ {
+ hvm_unmap_ioreq_page(s, IOREQ_PAGE_TYPE_IOREQ);
+ return rc;
+ }
+
+ rc = hvm_map_ioreq_page(s, IOREQ_PAGE_TYPE_VMPORT, vmport_ioreq_pfn);
+
+ if ( rc )
+ {
+ if ( bufioreq_pfn != INVALID_GFN )
+ hvm_unmap_ioreq_page(s, IOREQ_PAGE_TYPE_BUFIOREQ);
+ hvm_unmap_ioreq_page(s, IOREQ_PAGE_TYPE_IOREQ);
+ }
return rc;
}
@@ -759,6 +847,8 @@ static int hvm_ioreq_server_setup_pages(struct hvm_ioreq_server *s,
struct domain *d = s->domain;
unsigned long ioreq_pfn = INVALID_GFN;
unsigned long bufioreq_pfn = INVALID_GFN;
+ unsigned long vmport_ioreq_pfn =
+ d->arch.hvm_domain.params[HVM_PARAM_VMPORT_REGS_PFN];
int rc;
if ( is_default )
@@ -770,7 +860,8 @@ static int hvm_ioreq_server_setup_pages(struct hvm_ioreq_server *s,
ASSERT(handle_bufioreq);
return hvm_ioreq_server_map_pages(s,
d->arch.hvm_domain.params[HVM_PARAM_IOREQ_PFN],
- d->arch.hvm_domain.params[HVM_PARAM_BUFIOREQ_PFN]);
+ d->arch.hvm_domain.params[HVM_PARAM_BUFIOREQ_PFN],
+ vmport_ioreq_pfn);
}
rc = hvm_alloc_ioreq_gmfn(d, &ioreq_pfn);
@@ -779,8 +870,8 @@ static int hvm_ioreq_server_setup_pages(struct hvm_ioreq_server *s,
rc = hvm_alloc_ioreq_gmfn(d, &bufioreq_pfn);
if ( !rc )
- rc = hvm_ioreq_server_map_pages(s, ioreq_pfn, bufioreq_pfn);
-
+ rc = hvm_ioreq_server_map_pages(s, ioreq_pfn, bufioreq_pfn,
+ vmport_ioreq_pfn);
if ( rc )
{
hvm_free_ioreq_gmfn(d, ioreq_pfn);
@@ -795,11 +886,15 @@ static void hvm_ioreq_server_unmap_pages(struct hvm_ioreq_server *s,
{
struct domain *d = s->domain;
bool_t handle_bufioreq = ( s->bufioreq.va != NULL );
+ bool_t handle_vmport_ioreq = ( s->vmport_ioreq.va != NULL );
+
+ if ( handle_vmport_ioreq )
+ hvm_unmap_ioreq_page(s, IOREQ_PAGE_TYPE_VMPORT);
if ( handle_bufioreq )
- hvm_unmap_ioreq_page(s, 1);
+ hvm_unmap_ioreq_page(s, IOREQ_PAGE_TYPE_BUFIOREQ);
- hvm_unmap_ioreq_page(s, 0);
+ hvm_unmap_ioreq_page(s, IOREQ_PAGE_TYPE_IOREQ);
if ( !is_default )
{
@@ -834,12 +929,38 @@ static int hvm_ioreq_server_alloc_rangesets(struct hvm_ioreq_server *s,
for ( i = 0; i < NR_IO_RANGE_TYPES; i++ )
{
char *name;
+ char *type_name = NULL;
+ unsigned int limit;
- rc = asprintf(&name, "ioreq_server %d %s", s->id,
- (i == HVMOP_IO_RANGE_PORT) ? "port" :
- (i == HVMOP_IO_RANGE_MEMORY) ? "memory" :
- (i == HVMOP_IO_RANGE_PCI) ? "pci" :
- "");
+ switch ( i )
+ {
+ case HVMOP_IO_RANGE_PORT:
+ type_name = "port";
+ limit = MAX_NR_IO_RANGES;
+ break;
+ case HVMOP_IO_RANGE_MEMORY:
+ type_name = "memory";
+ limit = MAX_NR_IO_RANGES;
+ break;
+ case HVMOP_IO_RANGE_PCI:
+ type_name = "pci";
+ limit = MAX_NR_IO_RANGES;
+ break;
+ case HVMOP_IO_RANGE_VMWARE_PORT:
+ type_name = "VMware port";
+ limit = 1;
+ break;
+ case HVMOP_IO_RANGE_TIMEOFFSET:
+ type_name = "timeoffset";
+ limit = 1;
+ break;
+ default:
+ break;
+ }
+ if ( !type_name )
+ continue;
+
+ rc = asprintf(&name, "ioreq_server %d %s", s->id, type_name);
if ( rc )
goto fail;
@@ -852,7 +973,12 @@ static int hvm_ioreq_server_alloc_rangesets(struct hvm_ioreq_server *s,
if ( !s->range[i] )
goto fail;
- rangeset_limit(s->range[i], MAX_NR_IO_RANGES);
+ rangeset_limit(s->range[i], limit);
+
+ /* VMware port */
+ if ( i == HVMOP_IO_RANGE_VMWARE_PORT &&
+ s->domain->arch.hvm_domain.is_vmware_port_enabled )
+ rc = rangeset_add_range(s->range[i], 1, 1);
}
done:
@@ -1157,6 +1283,8 @@ static int hvm_map_io_range_to_ioreq_server(struct domain *d, ioservid_t id,
case HVMOP_IO_RANGE_PORT:
case HVMOP_IO_RANGE_MEMORY:
case HVMOP_IO_RANGE_PCI:
+ case HVMOP_IO_RANGE_VMWARE_PORT:
+ case HVMOP_IO_RANGE_TIMEOFFSET:
r = s->range[type];
break;
@@ -1208,6 +1336,8 @@ static int hvm_unmap_io_range_from_ioreq_server(struct domain *d, ioservid_t id,
case HVMOP_IO_RANGE_PORT:
case HVMOP_IO_RANGE_MEMORY:
case HVMOP_IO_RANGE_PCI:
+ case HVMOP_IO_RANGE_VMWARE_PORT:
+ case HVMOP_IO_RANGE_TIMEOFFSET:
r = s->range[type];
break;
@@ -2425,9 +2555,6 @@ struct hvm_ioreq_server *hvm_select_ioreq_server(struct domain *d,
if ( list_empty(&d->arch.hvm_domain.ioreq_server.list) )
return NULL;
- if ( p->type != IOREQ_TYPE_COPY && p->type != IOREQ_TYPE_PIO )
- return d->arch.hvm_domain.default_ioreq_server;
-
cf8 = d->arch.hvm_domain.pci_cf8;
if ( p->type == IOREQ_TYPE_PIO &&
@@ -2480,7 +2607,10 @@ struct hvm_ioreq_server *hvm_select_ioreq_server(struct domain *d,
BUILD_BUG_ON(IOREQ_TYPE_PIO != HVMOP_IO_RANGE_PORT);
BUILD_BUG_ON(IOREQ_TYPE_COPY != HVMOP_IO_RANGE_MEMORY);
BUILD_BUG_ON(IOREQ_TYPE_PCI_CONFIG != HVMOP_IO_RANGE_PCI);
+ BUILD_BUG_ON(IOREQ_TYPE_VMWARE_PORT != HVMOP_IO_RANGE_VMWARE_PORT);
+ BUILD_BUG_ON(IOREQ_TYPE_TIMEOFFSET != HVMOP_IO_RANGE_TIMEOFFSET);
r = s->range[type];
+ ASSERT(r);
switch ( type )
{
@@ -2507,6 +2637,13 @@ struct hvm_ioreq_server *hvm_select_ioreq_server(struct domain *d,
}
break;
+ case IOREQ_TYPE_VMWARE_PORT:
+ case IOREQ_TYPE_TIMEOFFSET:
+ /* The 'special' range of [1,1] is checked for being enabled */
+ if ( rangeset_contains_singleton(r, 1) )
+ return s;
+
+ break;
}
}
@@ -2669,6 +2806,7 @@ void hvm_complete_assist_req(ioreq_t *p)
case IOREQ_TYPE_PCI_CONFIG:
ASSERT_UNREACHABLE();
break;
+ case IOREQ_TYPE_VMWARE_PORT:
case IOREQ_TYPE_COPY:
case IOREQ_TYPE_PIO:
if ( p->dir == IOREQ_READ )
diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
index c0964ec..c1379bf 100644
--- a/xen/arch/x86/hvm/io.c
+++ b/xen/arch/x86/hvm/io.c
@@ -192,6 +192,22 @@ void hvm_io_assist(ioreq_t *p)
(void)handle_mmio();
break;
case HVMIO_handle_pio_awaiting_completion:
+ if ( p->type == IOREQ_TYPE_VMWARE_PORT )
+ {
+ vmware_regs_t *vr = get_vmport_regs_any(NULL, curr);
+
+ if ( vr )
+ {
+ struct cpu_user_regs *regs = guest_cpu_user_regs();
+
+ /* Only change the 32bit part of the register */
+ regs->_ebx = vr->ebx;
+ regs->_ecx = vr->ecx;
+ regs->_edx = vr->edx;
+ regs->_esi = vr->esi;
+ regs->_edi = vr->edi;
+ }
+ }
if ( vio->io_size == 4 ) /* Needs zero extension. */
guest_cpu_user_regs()->rax = (uint32_t)p->data;
else
diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
index f24d8e3..5e14402 100644
--- a/xen/arch/x86/hvm/vmware/vmport.c
+++ b/xen/arch/x86/hvm/vmware/vmport.c
@@ -137,6 +137,19 @@ void vmport_register(struct domain *d)
register_portio_handler(d, BDOOR_PORT, 4, vmport_ioport);
}
+int vmport_check_port(unsigned int port, unsigned int bytes)
+{
+ struct vcpu *curr = current;
+ struct domain *currd = curr->domain;
+
+ if ( port + bytes > BDOOR_PORT && port < BDOOR_PORT + 4 &&
+ is_hvm_domain(currd) &&
+ currd->arch.hvm_domain.is_vmware_port_enabled ) {
+ return 0;
+ }
+ return 1;
+}
+
/*
* Local variables:
* mode: C
diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
index 726b977..a6c312f 100644
--- a/xen/include/asm-x86/hvm/domain.h
+++ b/xen/include/asm-x86/hvm/domain.h
@@ -48,7 +48,7 @@ struct hvm_ioreq_vcpu {
evtchn_port_t ioreq_evtchn;
};
-#define NR_IO_RANGE_TYPES (HVMOP_IO_RANGE_PCI + 1)
+#define NR_IO_RANGE_TYPES (HVMOP_IO_RANGE_VMWARE_PORT + 1)
#define MAX_NR_IO_RANGES 256
struct hvm_ioreq_server {
@@ -63,6 +63,7 @@ struct hvm_ioreq_server {
ioservid_t id;
struct hvm_ioreq_page ioreq;
struct list_head ioreq_vcpu_list;
+ struct hvm_ioreq_page vmport_ioreq;
struct hvm_ioreq_page bufioreq;
/* Lock to serialize access to buffered ioreq ring */
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 394c230..78752b4 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -523,6 +523,8 @@ extern bool_t opt_hvm_fep;
#endif
void vmport_register(struct domain *d);
+int vmport_check_port(unsigned int port, unsigned int bytes);
+vmware_regs_t *get_vmport_regs_any(struct hvm_ioreq_server *s, struct vcpu *v);
#endif /* __ASM_X86_HVM_HVM_H__ */
diff --git a/xen/include/public/hvm/hvm_op.h b/xen/include/public/hvm/hvm_op.h
index 9b84e84..4d10eee 100644
--- a/xen/include/public/hvm/hvm_op.h
+++ b/xen/include/public/hvm/hvm_op.h
@@ -321,6 +321,9 @@ DEFINE_XEN_GUEST_HANDLE(xen_hvm_get_ioreq_server_info_t);
*
* NOTE: unless an emulation request falls entirely within a range mapped
* by a secondary emulator, it will not be passed to that emulator.
+ *
+ * NOTE: The 'special' range of [1,1] is what is checked for on
+ * TIMEOFFSET and VMWARE_PORT.
*/
#define HVMOP_map_io_range_to_ioreq_server 19
#define HVMOP_unmap_io_range_from_ioreq_server 20
@@ -331,6 +334,8 @@ struct xen_hvm_io_range {
# define HVMOP_IO_RANGE_PORT 0 /* I/O port range */
# define HVMOP_IO_RANGE_MEMORY 1 /* MMIO range */
# define HVMOP_IO_RANGE_PCI 2 /* PCI segment/bus/dev/func range */
+# define HVMOP_IO_RANGE_TIMEOFFSET 7 /* TIMEOFFSET special range */
+# define HVMOP_IO_RANGE_VMWARE_PORT 9 /* VMware port special range */
uint64_aligned_t start, end; /* IN - inclusive start and end of range */
};
typedef struct xen_hvm_io_range xen_hvm_io_range_t;
diff --git a/xen/include/public/hvm/ioreq.h b/xen/include/public/hvm/ioreq.h
index 2e5809b..2f326cf 100644
--- a/xen/include/public/hvm/ioreq.h
+++ b/xen/include/public/hvm/ioreq.h
@@ -37,6 +37,7 @@
#define IOREQ_TYPE_PCI_CONFIG 2
#define IOREQ_TYPE_TIMEOFFSET 7
#define IOREQ_TYPE_INVALIDATE 8 /* mapcache */
+#define IOREQ_TYPE_VMWARE_PORT 9 /* pio + vmport registers */
/*
* VMExit dispatcher should cooperate with instruction decoder to
@@ -48,6 +49,8 @@
*
* 63....48|47..40|39..35|34..32|31........0
* SEGMENT |BUS |DEV |FN |OFFSET
+ *
+ * For I/O type IOREQ_TYPE_VMWARE_PORT also use the vmware_regs.
*/
struct ioreq {
uint64_t addr; /* physical address */
@@ -66,11 +69,25 @@ struct ioreq {
};
typedef struct ioreq ioreq_t;
+struct vmware_regs {
+ uint32_t esi;
+ uint32_t edi;
+ uint32_t ebx;
+ uint32_t ecx;
+ uint32_t edx;
+};
+typedef struct vmware_regs vmware_regs_t;
+
struct shared_iopage {
struct ioreq vcpu_ioreq[1];
};
typedef struct shared_iopage shared_iopage_t;
+struct shared_vmport_iopage {
+ struct vmware_regs vcpu_vmport_regs[1];
+};
+typedef struct shared_vmport_iopage shared_vmport_iopage_t;
+
struct buf_ioreq {
uint8_t type; /* I/O type */
uint8_t pad:1;
diff --git a/xen/include/public/hvm/params.h b/xen/include/public/hvm/params.h
index 7c73089..130eba9 100644
--- a/xen/include/public/hvm/params.h
+++ b/xen/include/public/hvm/params.h
@@ -50,6 +50,8 @@
#define HVM_PARAM_PAE_ENABLED 4
#define HVM_PARAM_IOREQ_PFN 5
+/* Extra vmport PFN. */
+#define HVM_PARAM_VMPORT_REGS_PFN 35
#define HVM_PARAM_BUFIOREQ_PFN 6
#define HVM_PARAM_BUFIOREQ_EVTCHN 26
@@ -187,6 +189,6 @@
/* Location of the VM Generation ID in guest physical address space. */
#define HVM_PARAM_VM_GENERATION_ID_ADDR 34
-#define HVM_NR_PARAMS 35
+#define HVM_NR_PARAMS 36
#endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* [PATCH v12 8/8] Add xentrace to vmware_port
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
` (6 preceding siblings ...)
2015-06-27 23:27 ` [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT Don Slutz
@ 2015-06-27 23:27 ` Don Slutz
2015-06-29 14:54 ` Andrew Cooper
2015-07-10 8:40 ` [PATCH v12 0/8] Xen VMware tools support Jan Beulich
8 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-06-27 23:27 UTC (permalink / raw)
To: xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Don Slutz,
Tim Deegan, George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
From: Don Slutz <dslutz@verizon.com>
Also added missing TRAP_DEBUG & VLAPIC.
Signed-off-by: Don Slutz <dslutz@verizon.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
CC: Don Slutz <don.slutz@gmail.com>
---
v12:
Switch VMPORT_IGNORED to port, regs->_eax.
v11:
No change
v10:
Added Acked-by: Ian Campbell
Added back in the trace point calls.
Why is cmd in this patch?
Because the trace points use it.
v9:
Dropped unneed VMPORT_UNHANDLED, VMPORT_DECODE.
v7:
Dropped some of the new traces.
Added HVMTRACE_ND7.
v6:
Dropped the attempt to use svm_nextrip_insn_length via
__get_instruction_length (added in v2). Just always look
at upto 15 bytes on AMD.
v5:
exitinfo1 is used twice.
Fixed.
tools/xentrace/formats | 5 +++++
xen/arch/x86/hvm/io.c | 3 +++
xen/arch/x86/hvm/vmware/vmport.c | 15 ++++++++++++---
xen/include/asm-x86/hvm/trace.h | 22 ++++++++++++++++++++++
xen/include/public/trace.h | 3 +++
5 files changed, 45 insertions(+), 3 deletions(-)
diff --git a/tools/xentrace/formats b/tools/xentrace/formats
index 5d7b72a..ac8800e 100644
--- a/tools/xentrace/formats
+++ b/tools/xentrace/formats
@@ -79,6 +79,11 @@
0x00082020 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) INTR_WINDOW [ value = 0x%(1)08x ]
0x00082021 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) NPF [ gpa = 0x%(2)08x%(1)08x mfn = 0x%(4)08x%(3)08x qual = 0x%(5)04x p2mt = 0x%(6)04x ]
0x00082023 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) TRAP [ vector = 0x%(1)02x ]
+0x00082024 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) TRAP_DEBUG [ exit_qualification = 0x%(1)08x ]
+0x00082025 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VLAPIC
+0x00082026 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_HANDLED [ cmd = %(1)d eax = 0x%(2)08x ebx = 0x%(3)08x ecx = 0x%(4)08x edx = 0x%(5)08x esi = 0x%(6)08x edi = 0x%(7)08x ]
+0x00082027 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_IGNORED [ port = %(1)d eax = 0x%(2)08x ]
+0x00082028 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_QEMU [ eax = 0x%(1)08x ebx = 0x%(2)08x ecx = 0x%(3)08x edx = 0x%(4)08x esi = 0x%(5)08x edi = 0x%(6)08x ]
0x0010f001 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) page_grant_map [ domid = %(1)d ]
0x0010f002 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) page_grant_unmap [ domid = %(1)d ]
diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
index c1379bf..e2d947b 100644
--- a/xen/arch/x86/hvm/io.c
+++ b/xen/arch/x86/hvm/io.c
@@ -206,6 +206,9 @@ void hvm_io_assist(ioreq_t *p)
regs->_edx = vr->edx;
regs->_esi = vr->esi;
regs->_edi = vr->edi;
+ HVMTRACE_ND(VMPORT_QEMU, 0, 1/*cycles*/, 6,
+ p->data, regs->_ebx, regs->_ecx,
+ regs->_edx, regs->_esi, regs->_edi);
}
}
if ( vio->io_size == 4 ) /* Needs zero extension. */
diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
index 5e14402..de29e2f 100644
--- a/xen/arch/x86/hvm/vmware/vmport.c
+++ b/xen/arch/x86/hvm/vmware/vmport.c
@@ -16,6 +16,7 @@
#include <xen/lib.h>
#include <asm/hvm/hvm.h>
#include <asm/hvm/support.h>
+#include <asm/hvm/trace.h>
#include "backdoor_def.h"
@@ -35,6 +36,7 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
if ( port == BDOOR_PORT && regs->_eax == BDOOR_MAGIC )
{
uint32_t new_eax = ~0u;
+ uint16_t cmd = regs->_ecx;
uint64_t value;
struct vcpu *curr = current;
struct domain *currd = curr->domain;
@@ -45,7 +47,7 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
* leaving the high 32-bits unchanged, unlike what one would
* expect to happen.
*/
- switch ( regs->_ecx & 0xffff )
+ switch ( cmd )
{
case BDOOR_CMD_GETMHZ:
new_eax = currd->arch.tsc_khz / 1000;
@@ -123,11 +125,18 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
/* Let backing DM handle */
return X86EMUL_UNHANDLEABLE;
}
+ HVMTRACE_ND7(VMPORT_HANDLED, 0, 0/*cycles*/, 7,
+ cmd, new_eax, regs->_ebx, regs->_ecx,
+ regs->_edx, regs->_esi, regs->_edi);
if ( dir == IOREQ_READ )
*val = new_eax;
}
- else if ( dir == IOREQ_READ )
- *val = ~0u;
+ else
+ {
+ HVMTRACE_2D(VMPORT_IGNORED, port, regs->_eax);
+ if ( dir == IOREQ_READ )
+ *val = ~0u;
+ }
return X86EMUL_OKAY;
}
diff --git a/xen/include/asm-x86/hvm/trace.h b/xen/include/asm-x86/hvm/trace.h
index de802a6..0ad805f 100644
--- a/xen/include/asm-x86/hvm/trace.h
+++ b/xen/include/asm-x86/hvm/trace.h
@@ -54,6 +54,9 @@
#define DO_TRC_HVM_TRAP DEFAULT_HVM_MISC
#define DO_TRC_HVM_TRAP_DEBUG DEFAULT_HVM_MISC
#define DO_TRC_HVM_VLAPIC DEFAULT_HVM_MISC
+#define DO_TRC_HVM_VMPORT_HANDLED DEFAULT_HVM_IO
+#define DO_TRC_HVM_VMPORT_IGNORED DEFAULT_HVM_IO
+#define DO_TRC_HVM_VMPORT_QEMU DEFAULT_HVM_IO
#define TRC_PAR_LONG(par) ((par)&0xFFFFFFFF),((par)>>32)
@@ -83,6 +86,25 @@
} \
} while(0)
+#define HVMTRACE_ND7(evt, modifier, cycles, count, d1, d2, d3, d4, d5, d6, d7) \
+ do { \
+ if ( unlikely(tb_init_done) && DO_TRC_HVM_ ## evt ) \
+ { \
+ struct { \
+ u32 d[7]; \
+ } _d; \
+ _d.d[0]=(d1); \
+ _d.d[1]=(d2); \
+ _d.d[2]=(d3); \
+ _d.d[3]=(d4); \
+ _d.d[4]=(d5); \
+ _d.d[5]=(d6); \
+ _d.d[6]=(d7); \
+ __trace_var(TRC_HVM_ ## evt | (modifier), cycles, \
+ sizeof(*_d.d) * count, &_d); \
+ } \
+ } while(0)
+
#define HVMTRACE_6D(evt, d1, d2, d3, d4, d5, d6) \
HVMTRACE_ND(evt, 0, 0, 6, d1, d2, d3, d4, d5, d6)
#define HVMTRACE_5D(evt, d1, d2, d3, d4, d5) \
diff --git a/xen/include/public/trace.h b/xen/include/public/trace.h
index 5211ae7..16b87f9 100644
--- a/xen/include/public/trace.h
+++ b/xen/include/public/trace.h
@@ -227,6 +227,9 @@
#define TRC_HVM_TRAP (TRC_HVM_HANDLER + 0x23)
#define TRC_HVM_TRAP_DEBUG (TRC_HVM_HANDLER + 0x24)
#define TRC_HVM_VLAPIC (TRC_HVM_HANDLER + 0x25)
+#define TRC_HVM_VMPORT_HANDLED (TRC_HVM_HANDLER + 0x26)
+#define TRC_HVM_VMPORT_IGNORED (TRC_HVM_HANDLER + 0x27)
+#define TRC_HVM_VMPORT_QEMU (TRC_HVM_HANDLER + 0x28)
#define TRC_HVM_IOPORT_WRITE (TRC_HVM_HANDLER + 0x216)
#define TRC_HVM_IOMEM_WRITE (TRC_HVM_HANDLER + 0x217)
--
1.8.3.1
^ permalink raw reply related [flat|nested] 29+ messages in thread
* Re: [PATCH v12 8/8] Add xentrace to vmware_port
2015-06-27 23:27 ` [PATCH v12 8/8] Add xentrace to vmware_port Don Slutz
@ 2015-06-29 14:54 ` Andrew Cooper
2015-06-29 16:11 ` Don Slutz
0 siblings, 1 reply; 29+ messages in thread
From: Andrew Cooper @ 2015-06-29 14:54 UTC (permalink / raw)
To: Don Slutz, xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Tim Deegan,
George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Boris Ostrovsky, Suravee Suthikulpanit
On 28/06/15 00:27, Don Slutz wrote:
> From: Don Slutz <dslutz@verizon.com>
>
> Also added missing TRAP_DEBUG & VLAPIC.
>
> Signed-off-by: Don Slutz <dslutz@verizon.com>
> Acked-by: Ian Campbell <ian.campbell@citrix.com>
> CC: Don Slutz <don.slutz@gmail.com>
> ---
> v12:
> Switch VMPORT_IGNORED to port, regs->_eax.
>
> v11:
> No change
>
> v10:
> Added Acked-by: Ian Campbell
> Added back in the trace point calls.
>
> Why is cmd in this patch?
> Because the trace points use it.
>
> v9:
> Dropped unneed VMPORT_UNHANDLED, VMPORT_DECODE.
>
> v7:
> Dropped some of the new traces.
> Added HVMTRACE_ND7.
>
> v6:
> Dropped the attempt to use svm_nextrip_insn_length via
> __get_instruction_length (added in v2). Just always look
> at upto 15 bytes on AMD.
>
> v5:
> exitinfo1 is used twice.
> Fixed.
>
> tools/xentrace/formats | 5 +++++
> xen/arch/x86/hvm/io.c | 3 +++
> xen/arch/x86/hvm/vmware/vmport.c | 15 ++++++++++++---
> xen/include/asm-x86/hvm/trace.h | 22 ++++++++++++++++++++++
> xen/include/public/trace.h | 3 +++
> 5 files changed, 45 insertions(+), 3 deletions(-)
>
> diff --git a/tools/xentrace/formats b/tools/xentrace/formats
> index 5d7b72a..ac8800e 100644
> --- a/tools/xentrace/formats
> +++ b/tools/xentrace/formats
> @@ -79,6 +79,11 @@
> 0x00082020 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) INTR_WINDOW [ value = 0x%(1)08x ]
> 0x00082021 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) NPF [ gpa = 0x%(2)08x%(1)08x mfn = 0x%(4)08x%(3)08x qual = 0x%(5)04x p2mt = 0x%(6)04x ]
> 0x00082023 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) TRAP [ vector = 0x%(1)02x ]
> +0x00082024 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) TRAP_DEBUG [ exit_qualification = 0x%(1)08x ]
> +0x00082025 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VLAPIC
> +0x00082026 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_HANDLED [ cmd = %(1)d eax = 0x%(2)08x ebx = 0x%(3)08x ecx = 0x%(4)08x edx = 0x%(5)08x esi = 0x%(6)08x edi = 0x%(7)08x ]
> +0x00082027 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_IGNORED [ port = %(1)d eax = 0x%(2)08x ]
> +0x00082028 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_QEMU [ eax = 0x%(1)08x ebx = 0x%(2)08x ecx = 0x%(3)08x edx = 0x%(4)08x esi = 0x%(5)08x edi = 0x%(6)08x ]
>
> 0x0010f001 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) page_grant_map [ domid = %(1)d ]
> 0x0010f002 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) page_grant_unmap [ domid = %(1)d ]
> diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
> index c1379bf..e2d947b 100644
> --- a/xen/arch/x86/hvm/io.c
> +++ b/xen/arch/x86/hvm/io.c
> @@ -206,6 +206,9 @@ void hvm_io_assist(ioreq_t *p)
> regs->_edx = vr->edx;
> regs->_esi = vr->esi;
> regs->_edi = vr->edi;
> + HVMTRACE_ND(VMPORT_QEMU, 0, 1/*cycles*/, 6,
> + p->data, regs->_ebx, regs->_ecx,
> + regs->_edx, regs->_esi, regs->_edi);
> }
> }
> if ( vio->io_size == 4 ) /* Needs zero extension. */
> diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
> index 5e14402..de29e2f 100644
> --- a/xen/arch/x86/hvm/vmware/vmport.c
> +++ b/xen/arch/x86/hvm/vmware/vmport.c
> @@ -16,6 +16,7 @@
> #include <xen/lib.h>
> #include <asm/hvm/hvm.h>
> #include <asm/hvm/support.h>
> +#include <asm/hvm/trace.h>
>
> #include "backdoor_def.h"
>
> @@ -35,6 +36,7 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
> if ( port == BDOOR_PORT && regs->_eax == BDOOR_MAGIC )
> {
> uint32_t new_eax = ~0u;
> + uint16_t cmd = regs->_ecx;
> uint64_t value;
> struct vcpu *curr = current;
> struct domain *currd = curr->domain;
> @@ -45,7 +47,7 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
> * leaving the high 32-bits unchanged, unlike what one would
> * expect to happen.
> */
> - switch ( regs->_ecx & 0xffff )
> + switch ( cmd )
> {
> case BDOOR_CMD_GETMHZ:
> new_eax = currd->arch.tsc_khz / 1000;
> @@ -123,11 +125,18 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
> /* Let backing DM handle */
> return X86EMUL_UNHANDLEABLE;
> }
> + HVMTRACE_ND7(VMPORT_HANDLED, 0, 0/*cycles*/, 7,
> + cmd, new_eax, regs->_ebx, regs->_ecx,
> + regs->_edx, regs->_esi, regs->_edi);
> if ( dir == IOREQ_READ )
> *val = new_eax;
> }
> - else if ( dir == IOREQ_READ )
> - *val = ~0u;
> + else
> + {
> + HVMTRACE_2D(VMPORT_IGNORED, port, regs->_eax);
> + if ( dir == IOREQ_READ )
> + *val = ~0u;
> + }
>
> return X86EMUL_OKAY;
> }
> diff --git a/xen/include/asm-x86/hvm/trace.h b/xen/include/asm-x86/hvm/trace.h
> index de802a6..0ad805f 100644
> --- a/xen/include/asm-x86/hvm/trace.h
> +++ b/xen/include/asm-x86/hvm/trace.h
> @@ -54,6 +54,9 @@
> #define DO_TRC_HVM_TRAP DEFAULT_HVM_MISC
> #define DO_TRC_HVM_TRAP_DEBUG DEFAULT_HVM_MISC
> #define DO_TRC_HVM_VLAPIC DEFAULT_HVM_MISC
> +#define DO_TRC_HVM_VMPORT_HANDLED DEFAULT_HVM_IO
> +#define DO_TRC_HVM_VMPORT_IGNORED DEFAULT_HVM_IO
> +#define DO_TRC_HVM_VMPORT_QEMU DEFAULT_HVM_IO
>
>
> #define TRC_PAR_LONG(par) ((par)&0xFFFFFFFF),((par)>>32)
> @@ -83,6 +86,25 @@
> } \
> } while(0)
>
> +#define HVMTRACE_ND7(evt, modifier, cycles, count, d1, d2, d3, d4, d5, d6, d7) \
Please do this by extending the existing infrastructure rather than
special-casing 7 on the side. (i.e. extend ND to take 7 parameters, and
introduce HVMTRACE_7D)
> + do { \
> + if ( unlikely(tb_init_done) && DO_TRC_HVM_ ## evt ) \
> + { \
> + struct { \
> + u32 d[7]; \
> + } _d; \
= { d1, d2, d3, d4, d5, d6, d7 }
will be far shorter, linewise.
~Andrew
> + _d.d[0]=(d1); \
> + _d.d[1]=(d2); \
> + _d.d[2]=(d3); \
> + _d.d[3]=(d4); \
> + _d.d[4]=(d5); \
> + _d.d[5]=(d6); \
> + _d.d[6]=(d7); \
> + __trace_var(TRC_HVM_ ## evt | (modifier), cycles, \
> + sizeof(*_d.d) * count, &_d); \
> + } \
> + } while(0)
> +
> #define HVMTRACE_6D(evt, d1, d2, d3, d4, d5, d6) \
> HVMTRACE_ND(evt, 0, 0, 6, d1, d2, d3, d4, d5, d6)
> #define HVMTRACE_5D(evt, d1, d2, d3, d4, d5) \
> diff --git a/xen/include/public/trace.h b/xen/include/public/trace.h
> index 5211ae7..16b87f9 100644
> --- a/xen/include/public/trace.h
> +++ b/xen/include/public/trace.h
> @@ -227,6 +227,9 @@
> #define TRC_HVM_TRAP (TRC_HVM_HANDLER + 0x23)
> #define TRC_HVM_TRAP_DEBUG (TRC_HVM_HANDLER + 0x24)
> #define TRC_HVM_VLAPIC (TRC_HVM_HANDLER + 0x25)
> +#define TRC_HVM_VMPORT_HANDLED (TRC_HVM_HANDLER + 0x26)
> +#define TRC_HVM_VMPORT_IGNORED (TRC_HVM_HANDLER + 0x27)
> +#define TRC_HVM_VMPORT_QEMU (TRC_HVM_HANDLER + 0x28)
>
> #define TRC_HVM_IOPORT_WRITE (TRC_HVM_HANDLER + 0x216)
> #define TRC_HVM_IOMEM_WRITE (TRC_HVM_HANDLER + 0x217)
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 8/8] Add xentrace to vmware_port
2015-06-29 14:54 ` Andrew Cooper
@ 2015-06-29 16:11 ` Don Slutz
0 siblings, 0 replies; 29+ messages in thread
From: Don Slutz @ 2015-06-29 16:11 UTC (permalink / raw)
To: Andrew Cooper, xen-devel
Cc: Kevin Tian, Keir Fraser, Ian Campbell, Stefano Stabellini,
Jun Nakajima, Eddie Dong, Ian Jackson, Don Slutz, Tim Deegan,
George Dunlap, Aravind Gopalakrishnan, Jan Beulich,
Boris Ostrovsky, Suravee Suthikulpanit
On 06/29/15 10:54, Andrew Cooper wrote:
> On 28/06/15 00:27, Don Slutz wrote:
>> From: Don Slutz <dslutz@verizon.com>
>>
>> Also added missing TRAP_DEBUG & VLAPIC.
>>
>> Signed-off-by: Don Slutz <dslutz@verizon.com>
>> Acked-by: Ian Campbell <ian.campbell@citrix.com>
>> CC: Don Slutz <don.slutz@gmail.com>
>> ---
>> v12:
>> Switch VMPORT_IGNORED to port, regs->_eax.
>>
>> v11:
>> No change
>>
>> v10:
>> Added Acked-by: Ian Campbell
>> Added back in the trace point calls.
>>
>> Why is cmd in this patch?
>> Because the trace points use it.
>>
>> v9:
>> Dropped unneed VMPORT_UNHANDLED, VMPORT_DECODE.
>>
>> v7:
>> Dropped some of the new traces.
>> Added HVMTRACE_ND7.
>>
>> v6:
>> Dropped the attempt to use svm_nextrip_insn_length via
>> __get_instruction_length (added in v2). Just always look
>> at upto 15 bytes on AMD.
>>
>> v5:
>> exitinfo1 is used twice.
>> Fixed.
>>
>> tools/xentrace/formats | 5 +++++
>> xen/arch/x86/hvm/io.c | 3 +++
>> xen/arch/x86/hvm/vmware/vmport.c | 15 ++++++++++++---
>> xen/include/asm-x86/hvm/trace.h | 22 ++++++++++++++++++++++
>> xen/include/public/trace.h | 3 +++
>> 5 files changed, 45 insertions(+), 3 deletions(-)
>>
>> diff --git a/tools/xentrace/formats b/tools/xentrace/formats
>> index 5d7b72a..ac8800e 100644
>> --- a/tools/xentrace/formats
>> +++ b/tools/xentrace/formats
>> @@ -79,6 +79,11 @@
>> 0x00082020 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) INTR_WINDOW [ value = 0x%(1)08x ]
>> 0x00082021 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) NPF [ gpa = 0x%(2)08x%(1)08x mfn = 0x%(4)08x%(3)08x qual = 0x%(5)04x p2mt = 0x%(6)04x ]
>> 0x00082023 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) TRAP [ vector = 0x%(1)02x ]
>> +0x00082024 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) TRAP_DEBUG [ exit_qualification = 0x%(1)08x ]
>> +0x00082025 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VLAPIC
>> +0x00082026 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_HANDLED [ cmd = %(1)d eax = 0x%(2)08x ebx = 0x%(3)08x ecx = 0x%(4)08x edx = 0x%(5)08x esi = 0x%(6)08x edi = 0x%(7)08x ]
>> +0x00082027 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_IGNORED [ port = %(1)d eax = 0x%(2)08x ]
>> +0x00082028 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) VMPORT_QEMU [ eax = 0x%(1)08x ebx = 0x%(2)08x ecx = 0x%(3)08x edx = 0x%(4)08x esi = 0x%(5)08x edi = 0x%(6)08x ]
>>
>> 0x0010f001 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) page_grant_map [ domid = %(1)d ]
>> 0x0010f002 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) page_grant_unmap [ domid = %(1)d ]
>> diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
>> index c1379bf..e2d947b 100644
>> --- a/xen/arch/x86/hvm/io.c
>> +++ b/xen/arch/x86/hvm/io.c
>> @@ -206,6 +206,9 @@ void hvm_io_assist(ioreq_t *p)
>> regs->_edx = vr->edx;
>> regs->_esi = vr->esi;
>> regs->_edi = vr->edi;
>> + HVMTRACE_ND(VMPORT_QEMU, 0, 1/*cycles*/, 6,
>> + p->data, regs->_ebx, regs->_ecx,
>> + regs->_edx, regs->_esi, regs->_edi);
>> }
>> }
>> if ( vio->io_size == 4 ) /* Needs zero extension. */
>> diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
>> index 5e14402..de29e2f 100644
>> --- a/xen/arch/x86/hvm/vmware/vmport.c
>> +++ b/xen/arch/x86/hvm/vmware/vmport.c
>> @@ -16,6 +16,7 @@
>> #include <xen/lib.h>
>> #include <asm/hvm/hvm.h>
>> #include <asm/hvm/support.h>
>> +#include <asm/hvm/trace.h>
>>
>> #include "backdoor_def.h"
>>
>> @@ -35,6 +36,7 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
>> if ( port == BDOOR_PORT && regs->_eax == BDOOR_MAGIC )
>> {
>> uint32_t new_eax = ~0u;
>> + uint16_t cmd = regs->_ecx;
>> uint64_t value;
>> struct vcpu *curr = current;
>> struct domain *currd = curr->domain;
>> @@ -45,7 +47,7 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
>> * leaving the high 32-bits unchanged, unlike what one would
>> * expect to happen.
>> */
>> - switch ( regs->_ecx & 0xffff )
>> + switch ( cmd )
>> {
>> case BDOOR_CMD_GETMHZ:
>> new_eax = currd->arch.tsc_khz / 1000;
>> @@ -123,11 +125,18 @@ static int vmport_ioport(int dir, uint32_t port, uint32_t bytes, uint32_t *val)
>> /* Let backing DM handle */
>> return X86EMUL_UNHANDLEABLE;
>> }
>> + HVMTRACE_ND7(VMPORT_HANDLED, 0, 0/*cycles*/, 7,
>> + cmd, new_eax, regs->_ebx, regs->_ecx,
>> + regs->_edx, regs->_esi, regs->_edi);
>> if ( dir == IOREQ_READ )
>> *val = new_eax;
>> }
>> - else if ( dir == IOREQ_READ )
>> - *val = ~0u;
>> + else
>> + {
>> + HVMTRACE_2D(VMPORT_IGNORED, port, regs->_eax);
>> + if ( dir == IOREQ_READ )
>> + *val = ~0u;
>> + }
>>
>> return X86EMUL_OKAY;
>> }
>> diff --git a/xen/include/asm-x86/hvm/trace.h b/xen/include/asm-x86/hvm/trace.h
>> index de802a6..0ad805f 100644
>> --- a/xen/include/asm-x86/hvm/trace.h
>> +++ b/xen/include/asm-x86/hvm/trace.h
>> @@ -54,6 +54,9 @@
>> #define DO_TRC_HVM_TRAP DEFAULT_HVM_MISC
>> #define DO_TRC_HVM_TRAP_DEBUG DEFAULT_HVM_MISC
>> #define DO_TRC_HVM_VLAPIC DEFAULT_HVM_MISC
>> +#define DO_TRC_HVM_VMPORT_HANDLED DEFAULT_HVM_IO
>> +#define DO_TRC_HVM_VMPORT_IGNORED DEFAULT_HVM_IO
>> +#define DO_TRC_HVM_VMPORT_QEMU DEFAULT_HVM_IO
>>
>>
>> #define TRC_PAR_LONG(par) ((par)&0xFFFFFFFF),((par)>>32)
>> @@ -83,6 +86,25 @@
>> } \
>> } while(0)
>>
>> +#define HVMTRACE_ND7(evt, modifier, cycles, count, d1, d2, d3, d4, d5, d6, d7) \
> Please do this by extending the existing infrastructure rather than
> special-casing 7 on the side. (i.e. extend ND to take 7 parameters, and
> introduce HVMTRACE_7D)
>
>> + do { \
>> + if ( unlikely(tb_init_done) && DO_TRC_HVM_ ## evt ) \
>> + { \
>> + struct { \
>> + u32 d[7]; \
>> + } _d; \
> = { d1, d2, d3, d4, d5, d6, d7 }
>
> will be far shorter, linewise.
Ok, will do.
-Don Slutz
> ~Andrew
>
>> + _d.d[0]=(d1); \
>> + _d.d[1]=(d2); \
>> + _d.d[2]=(d3); \
>> + _d.d[3]=(d4); \
>> + _d.d[4]=(d5); \
>> + _d.d[5]=(d6); \
>> + _d.d[6]=(d7); \
>> + __trace_var(TRC_HVM_ ## evt | (modifier), cycles, \
>> + sizeof(*_d.d) * count, &_d); \
>> + } \
>> + } while(0)
>> +
>> #define HVMTRACE_6D(evt, d1, d2, d3, d4, d5, d6) \
>> HVMTRACE_ND(evt, 0, 0, 6, d1, d2, d3, d4, d5, d6)
>> #define HVMTRACE_5D(evt, d1, d2, d3, d4, d5) \
>> diff --git a/xen/include/public/trace.h b/xen/include/public/trace.h
>> index 5211ae7..16b87f9 100644
>> --- a/xen/include/public/trace.h
>> +++ b/xen/include/public/trace.h
>> @@ -227,6 +227,9 @@
>> #define TRC_HVM_TRAP (TRC_HVM_HANDLER + 0x23)
>> #define TRC_HVM_TRAP_DEBUG (TRC_HVM_HANDLER + 0x24)
>> #define TRC_HVM_VLAPIC (TRC_HVM_HANDLER + 0x25)
>> +#define TRC_HVM_VMPORT_HANDLED (TRC_HVM_HANDLER + 0x26)
>> +#define TRC_HVM_VMPORT_IGNORED (TRC_HVM_HANDLER + 0x27)
>> +#define TRC_HVM_VMPORT_QEMU (TRC_HVM_HANDLER + 0x28)
>>
>> #define TRC_HVM_IOPORT_WRITE (TRC_HVM_HANDLER + 0x216)
>> #define TRC_HVM_IOMEM_WRITE (TRC_HVM_HANDLER + 0x217)
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 2/8] xen: Add support for VMware cpuid leaves
2015-06-27 23:27 ` [PATCH v12 2/8] xen: Add support for VMware cpuid leaves Don Slutz
@ 2015-07-01 20:01 ` Konrad Rzeszutek Wilk
2015-07-03 22:03 ` Don Slutz
2015-07-07 15:56 ` Ian Campbell
1 sibling, 1 reply; 29+ messages in thread
From: Konrad Rzeszutek Wilk @ 2015-07-01 20:01 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
> diff --git a/tools/libxl/libxl_x86.c b/tools/libxl/libxl_x86.c
> index ed2bd38..651b338 100644
> --- a/tools/libxl/libxl_x86.c
> +++ b/tools/libxl/libxl_x86.c
> @@ -5,8 +5,8 @@ int libxl__arch_domain_prepare_config(libxl__gc *gc,
> libxl_domain_config *d_config,
> xc_domain_configuration_t *xc_config)
> {
> - /* No specific configuration right now */
> -
> + /* Note: will be changed in a later patch */
One usually say which patch. Also missing '.'.
> + xc_config->vmware_hwver = 0;
> return 0;
> }
>
... snip..
> diff --git a/xen/arch/x86/hvm/vmware/cpuid.c b/xen/arch/x86/hvm/vmware/cpuid.c
> new file mode 100644
> index 0000000..0dff36b
> --- /dev/null
> +++ b/xen/arch/x86/hvm/vmware/cpuid.c
> @@ -0,0 +1,77 @@
> +/*
> + * arch/x86/hvm/vmware/cpuid.c
> + *
> + * Copyright (C) 2012 Verizon Corporation
s/2012/2015/ ?
> + *
> + * This file is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License Version 2 (GPLv2)
> + * as published by the Free Software Foundation.
> + *
> + * This file 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 for more details. <http://www.gnu.org/licenses/>.
> + */
> +
> +#include <xen/sched.h>
> +
> +#include <asm/hvm/hvm.h>
> +#include <asm/hvm/vmware.h>
> +
> +/*
> + * VMware hardware version 7 defines some of these cpuid levels,
> + * below is a brief description about those.
> + *
> + * Leaf 0x40000000, Hypervisor CPUID information
> + * # EAX: The maximum input value for hypervisor CPUID info (0x40000010).
> + * # EBX, ECX, EDX: Hypervisor vendor ID signature. E.g. "VMwareVMware"
> + *
> + * Leaf 0x40000010, Timing information.
> + * # EAX: (Virtual) TSC frequency in kHz.
> + * # EBX: (Virtual) Bus (local apic timer) frequency in kHz.
> + * # ECX, EDX: RESERVED
> + */
> +
> +int cpuid_vmware_leaves(uint32_t idx, uint32_t *eax, uint32_t *ebx,
> + uint32_t *ecx, uint32_t *edx)
> +{
> + struct domain *d = current->domain;
> +
> + if ( !is_vmware_domain(d) ||
> + d->arch.hvm_domain.vmware_hwver < 7 )
> + return 0;
> +
> + switch ( idx - 0x40000000 )
> + {
> + case 0x0:
> + *eax = 0x40000010; /* Largest leaf */
> + *ebx = 0x61774d56; /* "VMwa" */
> + *ecx = 0x4d566572; /* "reVM" */
> + *edx = 0x65726177; /* "ware" */
> + break;
> +
> + case 0x10:
> + /* (Virtual) TSC frequency in kHz. */
> + *eax = d->arch.tsc_khz;
> + /* (Virtual) Bus (local apic timer) frequency in kHz. */
> + *ebx = 1000000ull / APIC_BUS_CYCLE_NS;
> + *ecx = 0; /* Reserved */
> + *edx = 0; /* Reserved */
> + break;
> +
> + default:
> + return 0;
> + }
> +
> + return 1;
> +}
> +
> +/*
> + * Local variables:
> + * mode: C
> + * c-file-style: "BSD"
> + * c-basic-offset: 4
> + * tab-width: 4
> + * indent-tabs-mode: nil
> + * End:
> + */
> diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
> index ac62f20..129be1c 100644
> --- a/xen/arch/x86/traps.c
> +++ b/xen/arch/x86/traps.c
> @@ -750,8 +750,12 @@ int cpuid_hypervisor_leaves( uint32_t idx, uint32_t sub_idx,
> uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx)
> {
> struct domain *currd = current->domain;
> - /* Optionally shift out of the way of Viridian architectural leaves. */
> - uint32_t base = is_viridian_domain(currd) ? 0x40000100 : 0x40000000;
> + /*
> + * Optionally shift out of the way of Viridian or VMware
> + * architectural leaves.
> + */
> + uint32_t base = is_viridian_domain(currd) || is_vmware_domain(currd) ?
> + 0x40000100 : 0x40000000;
> uint32_t limit, dummy;
>
> idx -= base;
> diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
> index ad68fcf..ada8aaa 100644
> --- a/xen/include/asm-x86/hvm/domain.h
> +++ b/xen/include/asm-x86/hvm/domain.h
> @@ -110,6 +110,9 @@ struct hvm_domain {
>
> uint64_t *params;
>
> + /* emulated VMware Hardware Version */
> + uint64_t vmware_hwver;
> +
> /* Memory ranges with pinned cache attributes. */
> struct list_head pinned_cacheattr_ranges;
>
> diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
> index 57f9605..a074afe 100644
> --- a/xen/include/asm-x86/hvm/hvm.h
> +++ b/xen/include/asm-x86/hvm/hvm.h
> @@ -356,6 +356,12 @@ static inline unsigned long hvm_get_shadow_gs_base(struct vcpu *v)
> #define has_viridian_time_ref_count(d) \
> (is_viridian_domain(d) && (viridian_feature_mask(d) & HVMPV_time_ref_count))
>
> +#define vmware_feature_mask(d) \
> + ((d)->arch.hvm_domain.vmware_hwver)
> +
> +#define is_vmware_domain(d) \
> + (is_hvm_domain(d) && vmware_feature_mask(d))
> +
> void hvm_hypervisor_cpuid_leaf(uint32_t sub_idx,
> uint32_t *eax, uint32_t *ebx,
> uint32_t *ecx, uint32_t *edx);
> diff --git a/xen/include/asm-x86/hvm/vmware.h b/xen/include/asm-x86/hvm/vmware.h
> new file mode 100644
> index 0000000..8390173
> --- /dev/null
> +++ b/xen/include/asm-x86/hvm/vmware.h
> @@ -0,0 +1,33 @@
> +/*
> + * asm-x86/hvm/vmware.h
> + *
> + * Copyright (C) 2012 Verizon Corporation
s/2012/2015
> + *
> + * This file is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License Version 2 (GPLv2)
> + * as published by the Free Software Foundation.
> + *
> + * This file 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 for more details. <http://www.gnu.org/licenses/>.
> + */
> +
> +#ifndef ASM_X86_HVM_VMWARE_H__
> +#define ASM_X86_HVM_VMWARE_H__
> +
> +#include <xen/types.h>
> +
> +int cpuid_vmware_leaves(uint32_t idx, uint32_t *eax, uint32_t *ebx,
> + uint32_t *ecx, uint32_t *edx);
> +
> +#endif /* ASM_X86_HVM_VMWARE_H__ */
> +
> +/*
> + * Local variables:
> + * mode: C
> + * c-file-style: "BSD"
> + * c-basic-offset: 4
> + * indent-tabs-mode: nil
> + * End:
> + */
> diff --git a/xen/include/public/arch-x86/xen.h b/xen/include/public/arch-x86/xen.h
> index 2ecc9c9..f84d10d 100644
> --- a/xen/include/public/arch-x86/xen.h
> +++ b/xen/include/public/arch-x86/xen.h
> @@ -268,7 +268,7 @@ typedef struct arch_shared_info arch_shared_info_t;
> * XEN_DOMCTL_INTERFACE_VERSION.
> */
> struct xen_arch_domainconfig {
> - char dummy;
> + uint64_t vmware_hwver;
> };
> #endif
>
> --
> 1.8.3.1
>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 3/8] tools: Add vmware_hwver support
2015-06-27 23:27 ` [PATCH v12 3/8] tools: Add vmware_hwver support Don Slutz
@ 2015-07-01 20:14 ` Konrad Rzeszutek Wilk
2015-07-07 15:59 ` Ian Campbell
0 siblings, 1 reply; 29+ messages in thread
From: Konrad Rzeszutek Wilk @ 2015-07-01 20:14 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On Sat, Jun 27, 2015 at 07:27:40PM -0400, Don Slutz wrote:
> From: Don Slutz <dslutz@verizon.com>
>
> This is used to set xen_arch_domainconfig vmware_hw. It is set to
> the emulated VMware virtual hardware version.
>
> Currently 0, 3-4, 6-11 are good values. However the code only
> checks for == 0, != 0, or < 7.
>
> Signed-off-by: Don Slutz <dslutz@verizon.com>
> CC: Don Slutz <don.slutz@gmail.com>
> ---
> v12:
> I'm not sure this hunk has anything to do with this patch, nor
> what the semantic difference between the old and new text is
> supposed to be.
> Dropped comment change.
>
>
> v11:
> Dropped "If non-zero then default VGA to VMware's VGA"
>
> v10:
> LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE &
> LIBXL_HAVE_BUILDINFO_HVM_VMWARE_HWVER are arriving together
> a single umbrella could be used.
> Since I split the LIBXL_VGA_INTERFACE_TYPE_VMWARE into
> it's own patch, this is not longer true.
> But I did use 1 for the 2 c_info changes.
> Please use GCSPRINTF.
> Remove vga=vmware from here.
>
> v9:
> I assumed that s/vmware_hw/vmware_hwver/ is not a big enough
> change to drop the Reviewed-by. Did a minor edit to the
> commit message to add 7 to the list of values checked.
>
> v7:
> Default handling of hvm.vga.kind bad.
> Fixed.
> Default of vmware_port should be based on vmware_hw.
> Done.
>
> v5:
> Anything looking for Xen according to the Xen cpuid instructions...
> Adjusted doc to new wording.
>
> docs/man/xl.cfg.pod.5 | 17 +++++++++++++++++
> tools/libxl/libxl_create.c | 4 +++-
> tools/libxl/libxl_types.idl | 1 +
> tools/libxl/libxl_x86.c | 3 +--
> tools/libxl/xl_cmdimpl.c | 2 ++
> 5 files changed, 24 insertions(+), 3 deletions(-)
>
> diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
> index 84078f6..4a01527 100644
> --- a/docs/man/xl.cfg.pod.5
> +++ b/docs/man/xl.cfg.pod.5
> @@ -1348,6 +1348,23 @@ The viridian option can be specified as a boolean. A value of true (1)
> is equivalent to the list [ "defaults" ], and a value of false (0) is
> equivalent to an empty list.
>
> +=item B<vmware_hwver=NUMBER>
> +
> +Turns on or off the exposure of VMware cpuid. The number is
> +VMware's hardware version number, where 0 is off. A number >= 7
> +is needed to enable exposure of VMware cpuid.
> +
> +The hardware version number (vmware_hwver) comes from VMware config files.
> +
> +=over 4
> +
> +In a .vmx it is virtualHW.version
> +
> +In a .ovf it is part of the value of vssd:VirtualSystemType.
> +For vssd:VirtualSystemType == vmx-07, vmware_hwver = 7.
> +
> +=back
> +
Perhaps add 'Recommended value is 7' ?
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT
2015-06-27 23:27 ` [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT Don Slutz
@ 2015-07-01 20:49 ` Konrad Rzeszutek Wilk
2015-07-03 14:55 ` Don Slutz
0 siblings, 1 reply; 29+ messages in thread
From: Konrad Rzeszutek Wilk @ 2015-07-01 20:49 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On Sat, Jun 27, 2015 at 07:27:44PM -0400, Don Slutz wrote:
> From: Don Slutz <dslutz@verizon.com>
>
> This adds synchronization of the 6 vcpu registers (only 32bits of
> them) that vmport.c needs between Xen and QEMU.
>
> This is to avoid a 2nd and 3rd exchange between QEMU and Xen to
> fetch and put these 6 vcpu registers used by the code in vmport.c
> and vmmouse.c
>
> In the tools, enable usage of QEMU's vmport code.
>
> The currently most useful VMware port support that QEMU has is the
> VMware mouse support. Xorg included a VMware mouse support that
> uses absolute mode. This make using a mouse in X11 much nicer.
>
> Signed-off-by: Don Slutz <dslutz@verizon.com>
> Acked-by: Ian Campbell <ian.campbell@citrix.com>
> CC: Don Slutz <don.slutz@gmail.com>
> ---
> v12:
> Rebase changes.
>
> Pass size to vmport_check_port() -- required if overlap
> I.E. inl on port 0x5657, 0x5656, 0x5655, 0x5659, 0x565a,
> and 0x565b.
>
> Move define of vmport_check_port() into this patch from ring3
> patch.
>
> v11:
> No change
>
> v10:
> These literals should become an enum.
> I don't think the invalidate type is needed.
> Code handling "case X86EMUL_UNHANDLEABLE:" in emulate.c
> is unclear.
> Comment about "special' range of 1" is not clear.
>
>
> v9:
> New code was presented as an RFC before this.
>
> Paul Durrant sugested I add support for other IOREQ types
> to HVMOP_map_io_range_to_ioreq_server.
> I have done this.
>
> tools/libxc/xc_hvm_build_x86.c | 5 +-
> tools/libxl/libxl_dm.c | 2 +
> xen/arch/x86/hvm/emulate.c | 83 +++++++++++++++---
> xen/arch/x86/hvm/hvm.c | 182 ++++++++++++++++++++++++++++++++++-----
> xen/arch/x86/hvm/io.c | 16 ++++
> xen/arch/x86/hvm/vmware/vmport.c | 13 +++
> xen/include/asm-x86/hvm/domain.h | 3 +-
> xen/include/asm-x86/hvm/hvm.h | 2 +
> xen/include/public/hvm/hvm_op.h | 5 ++
> xen/include/public/hvm/ioreq.h | 17 ++++
> xen/include/public/hvm/params.h | 4 +-
> 11 files changed, 293 insertions(+), 39 deletions(-)
>
> diff --git a/tools/libxc/xc_hvm_build_x86.c b/tools/libxc/xc_hvm_build_x86.c
> index 003ea06..feebbf7 100644
> --- a/tools/libxc/xc_hvm_build_x86.c
> +++ b/tools/libxc/xc_hvm_build_x86.c
> @@ -46,7 +46,8 @@
> #define SPECIALPAGE_IOREQ 5
> #define SPECIALPAGE_IDENT_PT 6
> #define SPECIALPAGE_CONSOLE 7
> -#define NR_SPECIAL_PAGES 8
> +#define SPECIALPAGE_VMPORT_REGS 8
> +#define NR_SPECIAL_PAGES 9
> #define special_pfn(x) (0xff000u - NR_SPECIAL_PAGES + (x))
>
> #define NR_IOREQ_SERVER_PAGES 8
> @@ -610,6 +611,8 @@ static int setup_guest(xc_interface *xch,
> special_pfn(SPECIALPAGE_BUFIOREQ));
> xc_hvm_param_set(xch, dom, HVM_PARAM_IOREQ_PFN,
> special_pfn(SPECIALPAGE_IOREQ));
> + xc_hvm_param_set(xch, dom, HVM_PARAM_VMPORT_REGS_PFN,
> + special_pfn(SPECIALPAGE_VMPORT_REGS));
> xc_hvm_param_set(xch, dom, HVM_PARAM_CONSOLE_PFN,
> special_pfn(SPECIALPAGE_CONSOLE));
> xc_hvm_param_set(xch, dom, HVM_PARAM_PAGING_RING_PFN,
> diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
> index 975996d..966ca46 100644
> --- a/tools/libxl/libxl_dm.c
> +++ b/tools/libxl/libxl_dm.c
> @@ -831,6 +831,8 @@ static int libxl__build_device_model_args_new(libxl__gc *gc,
> machinearg, max_ram_below_4g);
> }
> }
> + if (libxl_defbool_val(c_info->vmware_port))
> + machinearg = GCSPRINTF("%s,vmport=on", machinearg);
> flexarray_append(dm_args, machinearg);
> for (i = 0; b_info->extra_hvm && b_info->extra_hvm[i] != NULL; i++)
> flexarray_append(dm_args, b_info->extra_hvm[i]);
> diff --git a/xen/arch/x86/hvm/emulate.c b/xen/arch/x86/hvm/emulate.c
> index fe5661d..51a97d5 100644
> --- a/xen/arch/x86/hvm/emulate.c
> +++ b/xen/arch/x86/hvm/emulate.c
> @@ -169,33 +169,88 @@ static int hvmemul_do_io(
> vio->io_state = HVMIO_handle_mmio_awaiting_completion;
> break;
> case X86EMUL_UNHANDLEABLE:
> - {
> - struct hvm_ioreq_server *s =
> - hvm_select_ioreq_server(curr->domain, &p);
> -
> - /* If there is no suitable backing DM, just ignore accesses */
> - if ( !s )
> + if ( vmport_check_port(p.addr, p.size) )
I would have made this !. Thought if Jan or Andrew said to make it that
way - then please ignore me.
That would mean you could also change the function to be 'is_port_vmport' or
such.
> {
> - hvm_complete_assist_req(&p);
> - rc = X86EMUL_OKAY;
> - vio->io_state = HVMIO_none;
> + struct hvm_ioreq_server *s =
> + hvm_select_ioreq_server(curr->domain, &p);
> +
> + /* If there is no suitable backing DM, just ignore accesses */
> + if ( !s )
> + {
> + hvm_complete_assist_req(&p);
> + rc = X86EMUL_OKAY;
> + vio->io_state = HVMIO_none;
> + }
> + else
> + {
> + rc = hvm_send_assist_req(s, &p);
> + if ( rc != X86EMUL_RETRY )
> + vio->io_state = HVMIO_none;
> + else if ( data_is_addr || dir == IOREQ_WRITE )
> + rc = X86EMUL_OKAY;
> + }
> }
> else
> {
> - rc = hvm_send_assist_req(s, &p);
> - if ( rc != X86EMUL_RETRY )
> - vio->io_state = HVMIO_none;
> - else if ( data_is_addr || dir == IOREQ_WRITE )
> + struct hvm_ioreq_server *s;
> + vmware_regs_t *vr;
> +
> + BUILD_BUG_ON(sizeof(ioreq_t) < sizeof(vmware_regs_t));
> +
> + p.type = IOREQ_TYPE_VMWARE_PORT;
> + s = hvm_select_ioreq_server(curr->domain, &p);
> + vr = get_vmport_regs_any(s, curr);
> +
> + /*
> + * If there is no suitable backing DM, just ignore accesses. If
> + * we do not have access to registers to pass to QEMU, just
> + * ignore access.
> + */
> + if ( !s || !vr )
> + {
> + hvm_complete_assist_req(&p);
> rc = X86EMUL_OKAY;
> + vio->io_state = HVMIO_none;
> + }
> + else
> + {
> + struct cpu_user_regs *regs = guest_cpu_user_regs();
> +
> + p.data = regs->rax;
> + vr->ebx = regs->_ebx;
> + vr->ecx = regs->_ecx;
> + vr->edx = regs->_edx;
> + vr->esi = regs->_esi;
> + vr->edi = regs->_edi;
> +
> + vio->io_state = HVMIO_handle_pio_awaiting_completion;
> + rc = hvm_send_assist_req(s, &p);
> + if ( rc != X86EMUL_RETRY )
> + {
> + vio->io_state = HVMIO_none;
> + if ( rc == X86EMUL_OKAY )
> + rc = X86EMUL_RETRY;
> + }
> + }
> }
> break;
> - }
> default:
> BUG();
> }
>
> if ( rc != X86EMUL_OKAY )
> + {
> + /*
> + * If rc is X86EMUL_RETRY and vio->io_state is
> + * HVMIO_handle_pio_awaiting_completion, then were are of
were are? Perhaps 'were'?
> + * type IOREQ_TYPE_VMWARE_PORT, so completion in
> + * hvm_io_assist() with no re-emulation required
.. is required.
> + */
> + if ( rc == X86EMUL_RETRY &&
> + vio->io_state == HVMIO_handle_pio_awaiting_completion )
> + rc = X86EMUL_OKAY;
> return rc;
> + }
>
> finish_access:
> if ( dir == IOREQ_READ )
> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> index f8ec170..ce8cd09 100644
> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -394,6 +394,47 @@ static ioreq_t *get_ioreq(struct hvm_ioreq_server *s, struct vcpu *v)
> return &p->vcpu_ioreq[v->vcpu_id];
> }
>
> +static vmware_regs_t *get_vmport_regs_one(struct hvm_ioreq_server *s,
> + struct vcpu *v)
> +{
> + struct hvm_ioreq_vcpu *sv;
> +
> + list_for_each_entry ( sv,
> + &s->ioreq_vcpu_list,
> + list_entry )
Could it be just made it one nice line?
.. snip..
> @@ -2507,6 +2637,13 @@ struct hvm_ioreq_server *hvm_select_ioreq_server(struct domain *d,
> }
>
> break;
> + case IOREQ_TYPE_VMWARE_PORT:
> + case IOREQ_TYPE_TIMEOFFSET:
> + /* The 'special' range of [1,1] is checked for being enabled */
Missing full stop.
> + if ( rangeset_contains_singleton(r, 1) )
> + return s;
> +
> + break;
> }
> }
>
> @@ -2669,6 +2806,7 @@ void hvm_complete_assist_req(ioreq_t *p)
> case IOREQ_TYPE_PCI_CONFIG:
> ASSERT_UNREACHABLE();
> break;
> + case IOREQ_TYPE_VMWARE_PORT:
> case IOREQ_TYPE_COPY:
> case IOREQ_TYPE_PIO:
> if ( p->dir == IOREQ_READ )
> diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
> index c0964ec..c1379bf 100644
> --- a/xen/arch/x86/hvm/io.c
> +++ b/xen/arch/x86/hvm/io.c
> @@ -192,6 +192,22 @@ void hvm_io_assist(ioreq_t *p)
> (void)handle_mmio();
> break;
> case HVMIO_handle_pio_awaiting_completion:
> + if ( p->type == IOREQ_TYPE_VMWARE_PORT )
> + {
> + vmware_regs_t *vr = get_vmport_regs_any(NULL, curr);
> +
> + if ( vr )
> + {
> + struct cpu_user_regs *regs = guest_cpu_user_regs();
> +
> + /* Only change the 32bit part of the register */
Missing full stop.
> + regs->_ebx = vr->ebx;
> + regs->_ecx = vr->ecx;
> + regs->_edx = vr->edx;
> + regs->_esi = vr->esi;
> + regs->_edi = vr->edi;
> + }
> + }
> if ( vio->io_size == 4 ) /* Needs zero extension. */
> guest_cpu_user_regs()->rax = (uint32_t)p->data;
> else
> diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
> index f24d8e3..5e14402 100644
> --- a/xen/arch/x86/hvm/vmware/vmport.c
> +++ b/xen/arch/x86/hvm/vmware/vmport.c
> @@ -137,6 +137,19 @@ void vmport_register(struct domain *d)
> register_portio_handler(d, BDOOR_PORT, 4, vmport_ioport);
> }
>
> +int vmport_check_port(unsigned int port, unsigned int bytes)
> +{
> + struct vcpu *curr = current;
> + struct domain *currd = curr->domain;
> +
> + if ( port + bytes > BDOOR_PORT && port < BDOOR_PORT + 4 &&
> + is_hvm_domain(currd) &&
> + currd->arch.hvm_domain.is_vmware_port_enabled ) {
The '{' should be on its own line.
> + return 0;
> + }
> + return 1;
> +}
> +
> /*
> * Local variables:
> * mode: C
> diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
> index 726b977..a6c312f 100644
> --- a/xen/include/asm-x86/hvm/domain.h
> +++ b/xen/include/asm-x86/hvm/domain.h
> @@ -48,7 +48,7 @@ struct hvm_ioreq_vcpu {
> evtchn_port_t ioreq_evtchn;
> };
>
> -#define NR_IO_RANGE_TYPES (HVMOP_IO_RANGE_PCI + 1)
> +#define NR_IO_RANGE_TYPES (HVMOP_IO_RANGE_VMWARE_PORT + 1)
> #define MAX_NR_IO_RANGES 256
>
> struct hvm_ioreq_server {
> @@ -63,6 +63,7 @@ struct hvm_ioreq_server {
> ioservid_t id;
> struct hvm_ioreq_page ioreq;
> struct list_head ioreq_vcpu_list;
> + struct hvm_ioreq_page vmport_ioreq;
> struct hvm_ioreq_page bufioreq;
>
> /* Lock to serialize access to buffered ioreq ring */
> diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
> index 394c230..78752b4 100644
> --- a/xen/include/asm-x86/hvm/hvm.h
> +++ b/xen/include/asm-x86/hvm/hvm.h
> @@ -523,6 +523,8 @@ extern bool_t opt_hvm_fep;
> #endif
>
> void vmport_register(struct domain *d);
> +int vmport_check_port(unsigned int port, unsigned int bytes);
> +vmware_regs_t *get_vmport_regs_any(struct hvm_ioreq_server *s, struct vcpu *v);
>
> #endif /* __ASM_X86_HVM_HVM_H__ */
>
> diff --git a/xen/include/public/hvm/hvm_op.h b/xen/include/public/hvm/hvm_op.h
> index 9b84e84..4d10eee 100644
> --- a/xen/include/public/hvm/hvm_op.h
> +++ b/xen/include/public/hvm/hvm_op.h
> @@ -321,6 +321,9 @@ DEFINE_XEN_GUEST_HANDLE(xen_hvm_get_ioreq_server_info_t);
> *
> * NOTE: unless an emulation request falls entirely within a range mapped
> * by a secondary emulator, it will not be passed to that emulator.
> + *
> + * NOTE: The 'special' range of [1,1] is what is checked for on
> + * TIMEOFFSET and VMWARE_PORT.
> */
> #define HVMOP_map_io_range_to_ioreq_server 19
> #define HVMOP_unmap_io_range_from_ioreq_server 20
> @@ -331,6 +334,8 @@ struct xen_hvm_io_range {
> # define HVMOP_IO_RANGE_PORT 0 /* I/O port range */
> # define HVMOP_IO_RANGE_MEMORY 1 /* MMIO range */
> # define HVMOP_IO_RANGE_PCI 2 /* PCI segment/bus/dev/func range */
> +# define HVMOP_IO_RANGE_TIMEOFFSET 7 /* TIMEOFFSET special range */
> +# define HVMOP_IO_RANGE_VMWARE_PORT 9 /* VMware port special range */
> uint64_aligned_t start, end; /* IN - inclusive start and end of range */
> };
> typedef struct xen_hvm_io_range xen_hvm_io_range_t;
> diff --git a/xen/include/public/hvm/ioreq.h b/xen/include/public/hvm/ioreq.h
> index 2e5809b..2f326cf 100644
> --- a/xen/include/public/hvm/ioreq.h
> +++ b/xen/include/public/hvm/ioreq.h
> @@ -37,6 +37,7 @@
> #define IOREQ_TYPE_PCI_CONFIG 2
> #define IOREQ_TYPE_TIMEOFFSET 7
> #define IOREQ_TYPE_INVALIDATE 8 /* mapcache */
> +#define IOREQ_TYPE_VMWARE_PORT 9 /* pio + vmport registers */
>
> /*
> * VMExit dispatcher should cooperate with instruction decoder to
> @@ -48,6 +49,8 @@
> *
> * 63....48|47..40|39..35|34..32|31........0
> * SEGMENT |BUS |DEV |FN |OFFSET
> + *
> + * For I/O type IOREQ_TYPE_VMWARE_PORT also use the vmware_regs.
> */
> struct ioreq {
> uint64_t addr; /* physical address */
> @@ -66,11 +69,25 @@ struct ioreq {
> };
> typedef struct ioreq ioreq_t;
>
> +struct vmware_regs {
> + uint32_t esi;
> + uint32_t edi;
> + uint32_t ebx;
> + uint32_t ecx;
> + uint32_t edx;
> +};
> +typedef struct vmware_regs vmware_regs_t;
> +
> struct shared_iopage {
> struct ioreq vcpu_ioreq[1];
> };
> typedef struct shared_iopage shared_iopage_t;
>
> +struct shared_vmport_iopage {
> + struct vmware_regs vcpu_vmport_regs[1];
> +};
> +typedef struct shared_vmport_iopage shared_vmport_iopage_t;
> +
> struct buf_ioreq {
> uint8_t type; /* I/O type */
> uint8_t pad:1;
> diff --git a/xen/include/public/hvm/params.h b/xen/include/public/hvm/params.h
> index 7c73089..130eba9 100644
> --- a/xen/include/public/hvm/params.h
> +++ b/xen/include/public/hvm/params.h
> @@ -50,6 +50,8 @@
> #define HVM_PARAM_PAE_ENABLED 4
>
> #define HVM_PARAM_IOREQ_PFN 5
> +/* Extra vmport PFN. */
s/Extra/optional/ ?
> +#define HVM_PARAM_VMPORT_REGS_PFN 35
>
> #define HVM_PARAM_BUFIOREQ_PFN 6
> #define HVM_PARAM_BUFIOREQ_EVTCHN 26
> @@ -187,6 +189,6 @@
> /* Location of the VM Generation ID in guest physical address space. */
> #define HVM_PARAM_VM_GENERATION_ID_ADDR 34
>
> -#define HVM_NR_PARAMS 35
> +#define HVM_NR_PARAMS 36
>
> #endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */
> --
> 1.8.3.1
>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT
2015-07-01 20:49 ` Konrad Rzeszutek Wilk
@ 2015-07-03 14:55 ` Don Slutz
2015-07-07 15:02 ` Konrad Rzeszutek Wilk
0 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-07-03 14:55 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On 07/01/15 16:49, Konrad Rzeszutek Wilk wrote:
> On Sat, Jun 27, 2015 at 07:27:44PM -0400, Don Slutz wrote:
>> From: Don Slutz <dslutz@verizon.com>
>>
>> This adds synchronization of the 6 vcpu registers (only 32bits of
>> them) that vmport.c needs between Xen and QEMU.
>>
...
>> --- a/xen/arch/x86/hvm/emulate.c
>> +++ b/xen/arch/x86/hvm/emulate.c
>> @@ -169,33 +169,88 @@ static int hvmemul_do_io(
>> vio->io_state = HVMIO_handle_mmio_awaiting_completion;
>> break;
>> case X86EMUL_UNHANDLEABLE:
>> - {
>> - struct hvm_ioreq_server *s =
>> - hvm_select_ioreq_server(curr->domain, &p);
>> -
>> - /* If there is no suitable backing DM, just ignore accesses */
>> - if ( !s )
>> + if ( vmport_check_port(p.addr, p.size) )
> I would have made this !. Thought if Jan or Andrew said to make it that
> way - then please ignore me.
I have not checked, but I do not remember any direct statement either
way. Last was Jan's "looks better".
> That would mean you could also change the function to be 'is_port_vmport' or
> such.
>
Happy to make a change here if needed. Note: This part of the patch
conflicts with:
From: Paul Durrant <paul.durrant@citrix.com>
To: <xen-devel@lists.xenproject.org>
Date: Tue, 30 Jun 2015 14:05:43 +0100
Message-ID: <1435669558-5421-2-git-send-email-paul.durrant@citrix.com>
Subject: [Xen-devel] [PATCH v5 01/16] x86/hvm: make sure emulation is
retried if domain is shutting down
As so is waiting till that is sorted out.
>> {
>> - hvm_complete_assist_req(&p);
>> - rc = X86EMUL_OKAY;
>> - vio->io_state = HVMIO_none;
...
>> +
>> + vio->io_state = HVMIO_handle_pio_awaiting_completion;
>> + rc = hvm_send_assist_req(s, &p);
>> + if ( rc != X86EMUL_RETRY )
>> + {
>> + vio->io_state = HVMIO_none;
>> + if ( rc == X86EMUL_OKAY )
>> + rc = X86EMUL_RETRY;
>> + }
>> + }
>> }
>> break;
>> - }
>> default:
>> BUG();
>> }
>>
>> if ( rc != X86EMUL_OKAY )
>> + {
>> + /*
>> + * If rc is X86EMUL_RETRY and vio->io_state is
>> + * HVMIO_handle_pio_awaiting_completion, then were are of
> were are? Perhaps 'were'?
>> + * type IOREQ_TYPE_VMWARE_PORT, so completion in
>> + * hvm_io_assist() with no re-emulation required
> .. is required.
>
Ok. Plan to change.
With commit 2df1aa01 (and the result of Paul Durrant's change) I was
considering
just adding a "return X86EMUL_OKAY" as the else of the check on rc
above, and therefore drop the need for this check and comment.
>> + */
>> + if ( rc == X86EMUL_RETRY &&
>> + vio->io_state == HVMIO_handle_pio_awaiting_completion )
>> + rc = X86EMUL_OKAY;
>> return rc;
>> + }
>>
>> finish_access:
>> if ( dir == IOREQ_READ )
>> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
>> index f8ec170..ce8cd09 100644
>> --- a/xen/arch/x86/hvm/hvm.c
>> +++ b/xen/arch/x86/hvm/hvm.c
>> @@ -394,6 +394,47 @@ static ioreq_t *get_ioreq(struct hvm_ioreq_server *s, struct vcpu *v)
>> return &p->vcpu_ioreq[v->vcpu_id];
>> }
>>
>> +static vmware_regs_t *get_vmport_regs_one(struct hvm_ioreq_server *s,
>> + struct vcpu *v)
>> +{
>> + struct hvm_ioreq_vcpu *sv;
>> +
>> + list_for_each_entry ( sv,
>> + &s->ioreq_vcpu_list,
>> + list_entry )
> Could it be just made it one nice line?
I think so, so plan to change.
> .. snip..
>
>> @@ -2507,6 +2637,13 @@ struct hvm_ioreq_server *hvm_select_ioreq_server(struct domain *d,
>> }
>>
>> break;
>> + case IOREQ_TYPE_VMWARE_PORT:
>> + case IOREQ_TYPE_TIMEOFFSET:
>> + /* The 'special' range of [1,1] is checked for being enabled */
> Missing full stop.
Will add.
>> + if ( rangeset_contains_singleton(r, 1) )
>> + return s;
>> +
>> + break;
>> }
>> }
>>
>> @@ -2669,6 +2806,7 @@ void hvm_complete_assist_req(ioreq_t *p)
>> case IOREQ_TYPE_PCI_CONFIG:
>> ASSERT_UNREACHABLE();
>> break;
>> + case IOREQ_TYPE_VMWARE_PORT:
>> case IOREQ_TYPE_COPY:
>> case IOREQ_TYPE_PIO:
>> if ( p->dir == IOREQ_READ )
>> diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
>> index c0964ec..c1379bf 100644
>> --- a/xen/arch/x86/hvm/io.c
>> +++ b/xen/arch/x86/hvm/io.c
>> @@ -192,6 +192,22 @@ void hvm_io_assist(ioreq_t *p)
>> (void)handle_mmio();
>> break;
>> case HVMIO_handle_pio_awaiting_completion:
>> + if ( p->type == IOREQ_TYPE_VMWARE_PORT )
>> + {
>> + vmware_regs_t *vr = get_vmport_regs_any(NULL, curr);
>> +
>> + if ( vr )
>> + {
>> + struct cpu_user_regs *regs = guest_cpu_user_regs();
>> +
>> + /* Only change the 32bit part of the register */
> Missing full stop.
Will add.
>
>> + regs->_ebx = vr->ebx;
>> + regs->_ecx = vr->ecx;
>> + regs->_edx = vr->edx;
>> + regs->_esi = vr->esi;
>> + regs->_edi = vr->edi;
>> + }
>> + }
>> if ( vio->io_size == 4 ) /* Needs zero extension. */
>> guest_cpu_user_regs()->rax = (uint32_t)p->data;
>> else
>> diff --git a/xen/arch/x86/hvm/vmware/vmport.c b/xen/arch/x86/hvm/vmware/vmport.c
>> index f24d8e3..5e14402 100644
>> --- a/xen/arch/x86/hvm/vmware/vmport.c
>> +++ b/xen/arch/x86/hvm/vmware/vmport.c
>> @@ -137,6 +137,19 @@ void vmport_register(struct domain *d)
>> register_portio_handler(d, BDOOR_PORT, 4, vmport_ioport);
>> }
>>
>> +int vmport_check_port(unsigned int port, unsigned int bytes)
>> +{
>> + struct vcpu *curr = current;
>> + struct domain *currd = curr->domain;
>> +
>> + if ( port + bytes > BDOOR_PORT && port < BDOOR_PORT + 4 &&
>> + is_hvm_domain(currd) &&
>> + currd->arch.hvm_domain.is_vmware_port_enabled ) {
> The '{' should be on its own line.
Will fix.
>> + return 0;
>> + }
>> + return 1;
>> +}
>> +
...
>>
>> +struct shared_vmport_iopage {
>> + struct vmware_regs vcpu_vmport_regs[1];
>> +};
>> +typedef struct shared_vmport_iopage shared_vmport_iopage_t;
>> +
>> struct buf_ioreq {
>> uint8_t type; /* I/O type */
>> uint8_t pad:1;
>> diff --git a/xen/include/public/hvm/params.h b/xen/include/public/hvm/params.h
>> index 7c73089..130eba9 100644
>> --- a/xen/include/public/hvm/params.h
>> +++ b/xen/include/public/hvm/params.h
>> @@ -50,6 +50,8 @@
>> #define HVM_PARAM_PAE_ENABLED 4
>>
>> #define HVM_PARAM_IOREQ_PFN 5
>> +/* Extra vmport PFN. */
> s/Extra/optional/ ?
I do not think "optional" is good to use here. This PFN is required for
QEMU (or any other ioreq server) to handle vmport access. If you have a
better name then "extra", please let me know.
-Don Slutz
>> +#define HVM_PARAM_VMPORT_REGS_PFN 35
>>
>> #define HVM_PARAM_BUFIOREQ_PFN 6
>> #define HVM_PARAM_BUFIOREQ_EVTCHN 26
>> @@ -187,6 +189,6 @@
>> /* Location of the VM Generation ID in guest physical address space. */
>> #define HVM_PARAM_VM_GENERATION_ID_ADDR 34
>>
>> -#define HVM_NR_PARAMS 35
>> +#define HVM_NR_PARAMS 36
>>
>> #endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */
>> --
>> 1.8.3.1
>>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 2/8] xen: Add support for VMware cpuid leaves
2015-07-01 20:01 ` Konrad Rzeszutek Wilk
@ 2015-07-03 22:03 ` Don Slutz
2015-07-07 15:03 ` Konrad Rzeszutek Wilk
0 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-07-03 22:03 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On 07/01/15 16:01, Konrad Rzeszutek Wilk wrote:
>> diff --git a/tools/libxl/libxl_x86.c b/tools/libxl/libxl_x86.c
>> index ed2bd38..651b338 100644
>> --- a/tools/libxl/libxl_x86.c
>> +++ b/tools/libxl/libxl_x86.c
>> @@ -5,8 +5,8 @@ int libxl__arch_domain_prepare_config(libxl__gc *gc,
>> libxl_domain_config *d_config,
>> xc_domain_configuration_t *xc_config)
>> {
>> - /* No specific configuration right now */
>> -
>> + /* Note: will be changed in a later patch */
>
> One usually say which patch. Also missing '.'.
Ok, It is the next patch (tools: Add support for VMware cpuid leaves).
Will add the '.' and add a reference. This change was requested even
though it is fully replaced in the next patch.
>
>> + xc_config->vmware_hwver = 0;
>> return 0;
>> }
>>
> ... snip..
>
>> diff --git a/xen/arch/x86/hvm/vmware/cpuid.c b/xen/arch/x86/hvm/vmware/cpuid.c
>> new file mode 100644
>> index 0000000..0dff36b
>> --- /dev/null
>> +++ b/xen/arch/x86/hvm/vmware/cpuid.c
>> @@ -0,0 +1,77 @@
>> +/*
>> + * arch/x86/hvm/vmware/cpuid.c
>> + *
>> + * Copyright (C) 2012 Verizon Corporation
>
> s/2012/2015/ ?
The code here was written in 2012, 1st posted on xen-devel on Thu, 12
Dec 2013 14:15:11 -0500
([Xen-devel] [RFC PATCH 03/10] Add cpuid_vmware_leaves), and this same
code has been posted in 2014 and 2015.
So either no change or 2012-2015. Which do you want?
>> + *
>> + * This file is free software; you can redistribute it and/or modify it
>> + * under the terms of the GNU General Public License Version 2 (GPLv2)
>> + * as published by the Free Software Foundation.
>> + *
>> + * This file 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 for more details. <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#include <xen/sched.h>
>> +
>> +#include <asm/hvm/hvm.h>
>> +#include <asm/hvm/vmware.h>
>> +
>> +/*
>> + * VMware hardware version 7 defines some of these cpuid levels,
>> + * below is a brief description about those.
>> + *
>> + * Leaf 0x40000000, Hypervisor CPUID information
>> + * # EAX: The maximum input value for hypervisor CPUID info (0x40000010).
>> + * # EBX, ECX, EDX: Hypervisor vendor ID signature. E.g. "VMwareVMware"
>> + *
>> + * Leaf 0x40000010, Timing information.
>> + * # EAX: (Virtual) TSC frequency in kHz.
>> + * # EBX: (Virtual) Bus (local apic timer) frequency in kHz.
>> + * # ECX, EDX: RESERVED
>> + */
>> +
>> +int cpuid_vmware_leaves(uint32_t idx, uint32_t *eax, uint32_t *ebx,
>> + uint32_t *ecx, uint32_t *edx)
>> +{
>> + struct domain *d = current->domain;
>> +
>> + if ( !is_vmware_domain(d) ||
>> + d->arch.hvm_domain.vmware_hwver < 7 )
>> + return 0;
>> +
>> + switch ( idx - 0x40000000 )
>> + {
>> + case 0x0:
>> + *eax = 0x40000010; /* Largest leaf */
>> + *ebx = 0x61774d56; /* "VMwa" */
>> + *ecx = 0x4d566572; /* "reVM" */
>> + *edx = 0x65726177; /* "ware" */
>> + break;
>> +
>> + case 0x10:
>> + /* (Virtual) TSC frequency in kHz. */
>> + *eax = d->arch.tsc_khz;
>> + /* (Virtual) Bus (local apic timer) frequency in kHz. */
>> + *ebx = 1000000ull / APIC_BUS_CYCLE_NS;
>> + *ecx = 0; /* Reserved */
>> + *edx = 0; /* Reserved */
>> + break;
>> +
>> + default:
>> + return 0;
>> + }
>> +
>> + return 1;
>> +}
>> +
>> +/*
>> + * Local variables:
>> + * mode: C
>> + * c-file-style: "BSD"
>> + * c-basic-offset: 4
>> + * tab-width: 4
>> + * indent-tabs-mode: nil
>> + * End:
>> + */
>> diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
>> index ac62f20..129be1c 100644
>> --- a/xen/arch/x86/traps.c
>> +++ b/xen/arch/x86/traps.c
>> @@ -750,8 +750,12 @@ int cpuid_hypervisor_leaves( uint32_t idx, uint32_t sub_idx,
>> uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx)
>> {
>> struct domain *currd = current->domain;
>> - /* Optionally shift out of the way of Viridian architectural leaves. */
>> - uint32_t base = is_viridian_domain(currd) ? 0x40000100 : 0x40000000;
>> + /*
>> + * Optionally shift out of the way of Viridian or VMware
>> + * architectural leaves.
>> + */
>> + uint32_t base = is_viridian_domain(currd) || is_vmware_domain(currd) ?
>> + 0x40000100 : 0x40000000;
>> uint32_t limit, dummy;
>>
>> idx -= base;
>> diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
>> index ad68fcf..ada8aaa 100644
>> --- a/xen/include/asm-x86/hvm/domain.h
>> +++ b/xen/include/asm-x86/hvm/domain.h
>> @@ -110,6 +110,9 @@ struct hvm_domain {
>>
>> uint64_t *params;
>>
>> + /* emulated VMware Hardware Version */
>> + uint64_t vmware_hwver;
>> +
>> /* Memory ranges with pinned cache attributes. */
>> struct list_head pinned_cacheattr_ranges;
>>
>> diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
>> index 57f9605..a074afe 100644
>> --- a/xen/include/asm-x86/hvm/hvm.h
>> +++ b/xen/include/asm-x86/hvm/hvm.h
>> @@ -356,6 +356,12 @@ static inline unsigned long hvm_get_shadow_gs_base(struct vcpu *v)
>> #define has_viridian_time_ref_count(d) \
>> (is_viridian_domain(d) && (viridian_feature_mask(d) & HVMPV_time_ref_count))
>>
>> +#define vmware_feature_mask(d) \
>> + ((d)->arch.hvm_domain.vmware_hwver)
>> +
>> +#define is_vmware_domain(d) \
>> + (is_hvm_domain(d) && vmware_feature_mask(d))
>> +
>> void hvm_hypervisor_cpuid_leaf(uint32_t sub_idx,
>> uint32_t *eax, uint32_t *ebx,
>> uint32_t *ecx, uint32_t *edx);
>> diff --git a/xen/include/asm-x86/hvm/vmware.h b/xen/include/asm-x86/hvm/vmware.h
>> new file mode 100644
>> index 0000000..8390173
>> --- /dev/null
>> +++ b/xen/include/asm-x86/hvm/vmware.h
>> @@ -0,0 +1,33 @@
>> +/*
>> + * asm-x86/hvm/vmware.h
>> + *
>> + * Copyright (C) 2012 Verizon Corporation
>
> s/2012/2015
See above.
-Don Slutz
>> + *
>> + * This file is free software; you can redistribute it and/or modify it
>> + * under the terms of the GNU General Public License Version 2 (GPLv2)
>> + * as published by the Free Software Foundation.
>> + *
>> + * This file 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 for more details. <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#ifndef ASM_X86_HVM_VMWARE_H__
>> +#define ASM_X86_HVM_VMWARE_H__
>> +
>> +#include <xen/types.h>
>> +
>> +int cpuid_vmware_leaves(uint32_t idx, uint32_t *eax, uint32_t *ebx,
>> + uint32_t *ecx, uint32_t *edx);
>> +
>> +#endif /* ASM_X86_HVM_VMWARE_H__ */
>> +
>> +/*
>> + * Local variables:
>> + * mode: C
>> + * c-file-style: "BSD"
>> + * c-basic-offset: 4
>> + * indent-tabs-mode: nil
>> + * End:
>> + */
>> diff --git a/xen/include/public/arch-x86/xen.h b/xen/include/public/arch-x86/xen.h
>> index 2ecc9c9..f84d10d 100644
>> --- a/xen/include/public/arch-x86/xen.h
>> +++ b/xen/include/public/arch-x86/xen.h
>> @@ -268,7 +268,7 @@ typedef struct arch_shared_info arch_shared_info_t;
>> * XEN_DOMCTL_INTERFACE_VERSION.
>> */
>> struct xen_arch_domainconfig {
>> - char dummy;
>> + uint64_t vmware_hwver;
>> };
>> #endif
>>
>> --
>> 1.8.3.1
>>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT
2015-07-03 14:55 ` Don Slutz
@ 2015-07-07 15:02 ` Konrad Rzeszutek Wilk
2015-07-28 2:59 ` Don Slutz
0 siblings, 1 reply; 29+ messages in thread
From: Konrad Rzeszutek Wilk @ 2015-07-07 15:02 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
> >>diff --git a/xen/include/public/hvm/params.h b/xen/include/public/hvm/params.h
> >>index 7c73089..130eba9 100644
> >>--- a/xen/include/public/hvm/params.h
> >>+++ b/xen/include/public/hvm/params.h
> >>@@ -50,6 +50,8 @@
> >> #define HVM_PARAM_PAE_ENABLED 4
> >> #define HVM_PARAM_IOREQ_PFN 5
> >>+/* Extra vmport PFN. */
> >s/Extra/optional/ ?
>
> I do not think "optional" is good to use here. This PFN is required for
> QEMU (or any other ioreq server) to handle vmport access. If you have a
> better name then "extra", please let me know.
How about just 'VMWare Port PFN' ?
>
> -Don Slutz
>
> >>+#define HVM_PARAM_VMPORT_REGS_PFN 35
> >> #define HVM_PARAM_BUFIOREQ_PFN 6
> >> #define HVM_PARAM_BUFIOREQ_EVTCHN 26
> >>@@ -187,6 +189,6 @@
> >> /* Location of the VM Generation ID in guest physical address space. */
> >> #define HVM_PARAM_VM_GENERATION_ID_ADDR 34
> >>-#define HVM_NR_PARAMS 35
> >>+#define HVM_NR_PARAMS 36
> >> #endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */
> >>--
> >>1.8.3.1
> >>
>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 2/8] xen: Add support for VMware cpuid leaves
2015-07-03 22:03 ` Don Slutz
@ 2015-07-07 15:03 ` Konrad Rzeszutek Wilk
2015-07-28 2:54 ` Don Slutz
0 siblings, 1 reply; 29+ messages in thread
From: Konrad Rzeszutek Wilk @ 2015-07-07 15:03 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
> >> diff --git a/xen/arch/x86/hvm/vmware/cpuid.c b/xen/arch/x86/hvm/vmware/cpuid.c
> >> new file mode 100644
> >> index 0000000..0dff36b
> >> --- /dev/null
> >> +++ b/xen/arch/x86/hvm/vmware/cpuid.c
> >> @@ -0,0 +1,77 @@
> >> +/*
> >> + * arch/x86/hvm/vmware/cpuid.c
> >> + *
> >> + * Copyright (C) 2012 Verizon Corporation
> >
> > s/2012/2015/ ?
>
> The code here was written in 2012, 1st posted on xen-devel on Thu, 12
> Dec 2013 14:15:11 -0500
> ([Xen-devel] [RFC PATCH 03/10] Add cpuid_vmware_leaves), and this same
> code has been posted in 2014 and 2015.
>
> So either no change or 2012-2015. Which do you want?
It is up to you. Usually one posts with the latest copyright date.
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 1/8] tools: Add vga=vmware
2015-06-27 23:27 ` [PATCH v12 1/8] tools: Add vga=vmware Don Slutz
@ 2015-07-07 15:54 ` Ian Campbell
0 siblings, 0 replies; 29+ messages in thread
From: Ian Campbell @ 2015-07-07 15:54 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Jun Nakajima,
Stefano Stabellini, George Dunlap, Ian Jackson, Eddie Dong,
Don Slutz, xen-devel, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
On Sat, 2015-06-27 at 19:27 -0400, Don Slutz wrote:
> From: Don Slutz <dslutz@verizon.com>
>
> This allows use of QEMU's VMware emulated video card
>
> NOTE: vga=vmware is not supported by device_model_version=qemu-xen-traditional
>
> Signed-off-by: Don Slutz <dslutz@verizon.com>
> CC: Don Slutz <don.slutz@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
> ---
> v12:
> Dropped LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE
> This means that the later patch that defines LIBXL_HAVE_VMWARE
> is now also required.
>
> v11:
> Dropped support for Qemu-trad.
> Also changed later patchs to not need this one.
>
> v10: New at v10.
>
> Was part of "tools: Add vmware_hwver support"
>
> docs/man/xl.cfg.pod.5 | 4 +++-
> tools/libxl/libxl_dm.c | 9 +++++++++
> tools/libxl/libxl_types.idl | 1 +
> tools/libxl/xl_cmdimpl.c | 2 ++
> 4 files changed, 15 insertions(+), 1 deletion(-)
>
> diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
> index a3e0e2e..84078f6 100644
> --- a/docs/man/xl.cfg.pod.5
> +++ b/docs/man/xl.cfg.pod.5
> @@ -1389,7 +1389,7 @@ This option is deprecated, use vga="stdvga" instead.
>
> =item B<vga="STRING">
>
> -Selects the emulated video card (none|stdvga|cirrus|qxl).
> +Selects the emulated video card (none|stdvga|cirrus|qxl|vmware).
> The default is cirrus.
>
> In general, QXL should work with the Spice remote display protocol
> @@ -1397,6 +1397,8 @@ for acceleration, and QXL driver is necessary in guest in this case.
> QXL can also work with the VNC protocol, but it will be like a standard
> VGA without acceleration.
>
> +NOTE: vmware is not supported on B<device_model_version = "qemu-xen-traditional">
> +
> =item B<vnc=BOOLEAN>
>
> Allow access to the display via the VNC protocol. This enables the
> diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
> index 6b8b3ac..975996d 100644
> --- a/tools/libxl/libxl_dm.c
> +++ b/tools/libxl/libxl_dm.c
> @@ -256,6 +256,10 @@ static int libxl__build_device_model_args_old(libxl__gc *gc,
> case LIBXL_VGA_INTERFACE_TYPE_NONE:
> flexarray_append_pair(dm_args, "-vga", "none");
> break;
> + case LIBXL_VGA_INTERFACE_TYPE_VMWARE:
> + LOG(ERROR, "vga=vmware is not supported by "
> + "qemu-xen-traditional");
> + return ERROR_INVAL;
> case LIBXL_VGA_INTERFACE_TYPE_QXL:
> break;
> }
> @@ -657,6 +661,11 @@ static int libxl__build_device_model_args_new(libxl__gc *gc,
> GCSPRINTF("qxl-vga,vram_size_mb=%"PRIu64",ram_size_mb=%"PRIu64,
> (b_info->video_memkb/2/1024), (b_info->video_memkb/2/1024) ) );
> break;
> + case LIBXL_VGA_INTERFACE_TYPE_VMWARE:
> + flexarray_append_pair(dm_args, "-device",
> + GCSPRINTF("vmware-svga,vgamem_mb=%d",
> + libxl__sizekb_to_mb(b_info->video_memkb)));
> + break;
> }
>
> if (b_info->u.hvm.boot) {
> diff --git a/tools/libxl/libxl_types.idl b/tools/libxl/libxl_types.idl
> index 736c4e4..534e7f0 100644
> --- a/tools/libxl/libxl_types.idl
> +++ b/tools/libxl/libxl_types.idl
> @@ -187,6 +187,7 @@ libxl_vga_interface_type = Enumeration("vga_interface_type", [
> (2, "STD"),
> (3, "NONE"),
> (4, "QXL"),
> + (5, "VMWARE"),
> ], init_val = "LIBXL_VGA_INTERFACE_TYPE_CIRRUS")
>
> libxl_vendor_device = Enumeration("vendor_device", [
> diff --git a/tools/libxl/xl_cmdimpl.c b/tools/libxl/xl_cmdimpl.c
> index 568fd21..efb3651 100644
> --- a/tools/libxl/xl_cmdimpl.c
> +++ b/tools/libxl/xl_cmdimpl.c
> @@ -2142,6 +2142,8 @@ skip_vfb:
> b_info->u.hvm.vga.kind = LIBXL_VGA_INTERFACE_TYPE_NONE;
> } else if (!strcmp(buf, "qxl")) {
> b_info->u.hvm.vga.kind = LIBXL_VGA_INTERFACE_TYPE_QXL;
> + } else if (!strcmp(buf, "vmware")) {
> + b_info->u.hvm.vga.kind = LIBXL_VGA_INTERFACE_TYPE_VMWARE;
> } else {
> fprintf(stderr, "Unknown vga \"%s\" specified\n", buf);
> exit(1);
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 2/8] xen: Add support for VMware cpuid leaves
2015-06-27 23:27 ` [PATCH v12 2/8] xen: Add support for VMware cpuid leaves Don Slutz
2015-07-01 20:01 ` Konrad Rzeszutek Wilk
@ 2015-07-07 15:56 ` Ian Campbell
1 sibling, 0 replies; 29+ messages in thread
From: Ian Campbell @ 2015-07-07 15:56 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Jun Nakajima,
Stefano Stabellini, George Dunlap, Ian Jackson, Eddie Dong,
Don Slutz, xen-devel, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
On Sat, 2015-06-27 at 19:27 -0400, Don Slutz wrote:
> tools/libxl/libxl_x86.c | 4 +-
Tools side only: Acked-by: Ian Campbell <ian.campbell@citrix.com>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 3/8] tools: Add vmware_hwver support
2015-07-01 20:14 ` Konrad Rzeszutek Wilk
@ 2015-07-07 15:59 ` Ian Campbell
2015-07-28 2:58 ` Don Slutz
0 siblings, 1 reply; 29+ messages in thread
From: Ian Campbell @ 2015-07-07 15:59 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Jun Nakajima,
Stefano Stabellini, George Dunlap, Eddie Dong, Ian Jackson,
Don Slutz, Don Slutz, xen-devel, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On Wed, 2015-07-01 at 16:14 -0400, Konrad Rzeszutek Wilk wrote:
> On Sat, Jun 27, 2015 at 07:27:40PM -0400, Don Slutz wrote:
> > From: Don Slutz <dslutz@verizon.com>
> >
> > This is used to set xen_arch_domainconfig vmware_hw. It is set to
> > the emulated VMware virtual hardware version.
> >
> > Currently 0, 3-4, 6-11 are good values. However the code only
> > checks for == 0, != 0, or < 7.
> >
> > Signed-off-by: Don Slutz <dslutz@verizon.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
> >
> > docs/man/xl.cfg.pod.5 | 17 +++++++++++++++++
> > tools/libxl/libxl_create.c | 4 +++-
> > tools/libxl/libxl_types.idl | 1 +
> > tools/libxl/libxl_x86.c | 3 +--
> > tools/libxl/xl_cmdimpl.c | 2 ++
> > 5 files changed, 24 insertions(+), 3 deletions(-)
> >
> > diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
> > index 84078f6..4a01527 100644
> > --- a/docs/man/xl.cfg.pod.5
> > +++ b/docs/man/xl.cfg.pod.5
> > @@ -1348,6 +1348,23 @@ The viridian option can be specified as a boolean. A value of true (1)
> > is equivalent to the list [ "defaults" ], and a value of false (0) is
> > equivalent to an empty list.
> >
> > +=item B<vmware_hwver=NUMBER>
> > +
> > +Turns on or off the exposure of VMware cpuid. The number is
> > +VMware's hardware version number, where 0 is off. A number >= 7
> > +is needed to enable exposure of VMware cpuid.
> > +
> > +The hardware version number (vmware_hwver) comes from VMware config files.
> > +
> > +=over 4
> > +
> > +In a .vmx it is virtualHW.version
> > +
> > +In a .ovf it is part of the value of vssd:VirtualSystemType.
> > +For vssd:VirtualSystemType == vmx-07, vmware_hwver = 7.
> > +
> > +=back
> > +
>
> Perhaps add 'Recommended value is 7' ?
Assuming that is a good recommendation then that might be a good
enhancement for a future patch. Or perhaps some sort of simple mapping
from VMware product versions/names to a suitable value to use, unless
that means having to think about trademarks etc.
Ian.
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 6/8] tools: Add vmware_port support
2015-06-27 23:27 ` [PATCH v12 6/8] tools: " Don Slutz
@ 2015-07-07 16:01 ` Ian Campbell
0 siblings, 0 replies; 29+ messages in thread
From: Ian Campbell @ 2015-07-07 16:01 UTC (permalink / raw)
To: Don Slutz
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Jun Nakajima,
Stefano Stabellini, George Dunlap, Ian Jackson, Eddie Dong,
Don Slutz, xen-devel, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
On Sat, 2015-06-27 at 19:27 -0400, Don Slutz wrote:
> This new libxl_domain_create_info field is used to set
> XEN_DOMCTL_CONFIG_VMWARE_PORT_MASK in the xc_domain_configuration_t
> for x86.
>
> In xen it is is_vmware_port_enabled.
>
> If is_vmware_port_enabled then
> enable a limited support of VMware's hyper-call.
>
> VMware's hyper-call is also known as VMware Backdoor I/O Port.
>
> if vmware_port is not specified in the config file, let
> "vmware_hwver != 0" be the default value. This means that only
> vmware_hwver = 7 needs to be specified to enable both features.
>
> vmware_hwver = 7 is special because that is what controls the
> enable of CPUID leaves for VMware (vmware_hwver >= 7).
>
> Note: vmware_port and nestedhvm cannot be specified at the
> same time.
>
> Signed-off-by: Don Slutz <dslutz@verizon.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 0/8] Xen VMware tools support
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
` (7 preceding siblings ...)
2015-06-27 23:27 ` [PATCH v12 8/8] Add xentrace to vmware_port Don Slutz
@ 2015-07-10 8:40 ` Jan Beulich
2015-07-10 9:07 ` Wei Liu
8 siblings, 1 reply; 29+ messages in thread
From: Jan Beulich @ 2015-07-10 8:40 UTC (permalink / raw)
To: Wei Liu, Don Slutz
Cc: Jun Nakajima, Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, George Dunlap, Andrew Cooper, Ian Jackson,
xen-devel, Eddie Dong, Aravind Gopalakrishnan,
Suravee Suthikulpanit, Boris Ostrovsky
>>> On 28.06.15 at 01:27, <don.slutz@gmail.com> wrote:
> Changes v11 to v12:
> Rebased on staging.
>
> Jan Beulich (xen: Add ring 3 vmware_port support)
> As there don't seem to be enough convincing arguments for this
> to be worthwhile, I'm going to drop this...
> Dropped the 2 ring3 patches.
>
> Ian Campbell (tools: Add vmware_hwver support):
> s/come/comes/
> Done
> I'm not sure this hunk has anything to do with this patch, nor
> what the semantic difference between the old and new text is
> supposed to be.
> Dropped comment change.
>
> Ian Campbell (multiple):
> In v11 this seems to have morphed into only
> LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE being provided, which
> is clearly not an appropriate umbrella #define.
>
> "#define LIBXL_HAVE_CREATEINFO_VMWARE 1"
> Lets just have a single one of these indicating support for
> vmware, it should be added at the end of the series after all
> the baseline vmware functionality is in place. I think that
> means hwver, vga=vmware and this port stuff.
>
> Make (tools: Add vga=vmware) no longer independent.
> Change the #define to "LIBXL_HAVE_VMWARE"
>
> Jan Beulich (xen: Add vmware_port support):
> Surrounding code avoiding the use of "break" makes the result
> look rather inconsistent. Please move this up immediately after
> the XSM check, or drop the "break".
> Moved it up.
>
> George Dunlap (Add IOREQ_TYPE_VMWARE_PORT):
> Sorry for coming a bit late to this party. On a high level I
> think this is good, but there doesn't seem to be anything in
> here in particular that is vmware-specific. Would it make more
> sense to give this a more generic name, and have it include all
> of the general-purpose registers?
> Did not change at this time because QEMU would need changes 1st.
>
> George Dunlap (Add xentrace to vmware_port):
> Do you need to log edi as well? It looks like it's not used.
> I think it makes sense to indicate that edi can be used and
> may in the future.
> And do you need to log all the registers here? It seems like
> port + regs->_ecx would be enough to tell you why it got
> ignored.
> Adjusted to log just port and eax (ecx was NOT tested for).
I was actually meaning to apply this (at least until patch 6; I'm
not convinced patch 7 makes sense to go in at this point), but
noticed that comments by Konrad were still un-addressed. I'm
also expecting some re-basing to be needed on top of Paul's
emulation re-work.
Wei, considering that this had already missed 4.5, would you
be okay to take at least first 6 patches (which seem reasonably
risk free) after the freeze?
Jan
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 0/8] Xen VMware tools support
2015-07-10 8:40 ` [PATCH v12 0/8] Xen VMware tools support Jan Beulich
@ 2015-07-10 9:07 ` Wei Liu
2015-07-28 3:18 ` Don Slutz
0 siblings, 1 reply; 29+ messages in thread
From: Wei Liu @ 2015-07-10 9:07 UTC (permalink / raw)
To: Jan Beulich
Cc: Jun Nakajima, Tim Deegan, Kevin Tian, Wei Liu, Ian Campbell,
Stefano Stabellini, George Dunlap, Andrew Cooper, Ian Jackson,
Don Slutz, xen-devel, Eddie Dong, Aravind Gopalakrishnan,
Suravee Suthikulpanit, Keir Fraser, Boris Ostrovsky
On Fri, Jul 10, 2015 at 09:40:59AM +0100, Jan Beulich wrote:
> >>> On 28.06.15 at 01:27, <don.slutz@gmail.com> wrote:
> > Changes v11 to v12:
> > Rebased on staging.
> >
> > Jan Beulich (xen: Add ring 3 vmware_port support)
> > As there don't seem to be enough convincing arguments for this
> > to be worthwhile, I'm going to drop this...
> > Dropped the 2 ring3 patches.
> >
> > Ian Campbell (tools: Add vmware_hwver support):
> > s/come/comes/
> > Done
> > I'm not sure this hunk has anything to do with this patch, nor
> > what the semantic difference between the old and new text is
> > supposed to be.
> > Dropped comment change.
> >
> > Ian Campbell (multiple):
> > In v11 this seems to have morphed into only
> > LIBXL_HAVE_LIBXL_VGA_INTERFACE_TYPE_VMWARE being provided, which
> > is clearly not an appropriate umbrella #define.
> >
> > "#define LIBXL_HAVE_CREATEINFO_VMWARE 1"
> > Lets just have a single one of these indicating support for
> > vmware, it should be added at the end of the series after all
> > the baseline vmware functionality is in place. I think that
> > means hwver, vga=vmware and this port stuff.
> >
> > Make (tools: Add vga=vmware) no longer independent.
> > Change the #define to "LIBXL_HAVE_VMWARE"
> >
> > Jan Beulich (xen: Add vmware_port support):
> > Surrounding code avoiding the use of "break" makes the result
> > look rather inconsistent. Please move this up immediately after
> > the XSM check, or drop the "break".
> > Moved it up.
> >
> > George Dunlap (Add IOREQ_TYPE_VMWARE_PORT):
> > Sorry for coming a bit late to this party. On a high level I
> > think this is good, but there doesn't seem to be anything in
> > here in particular that is vmware-specific. Would it make more
> > sense to give this a more generic name, and have it include all
> > of the general-purpose registers?
> > Did not change at this time because QEMU would need changes 1st.
> >
> > George Dunlap (Add xentrace to vmware_port):
> > Do you need to log edi as well? It looks like it's not used.
> > I think it makes sense to indicate that edi can be used and
> > may in the future.
> > And do you need to log all the registers here? It seems like
> > port + regs->_ecx would be enough to tell you why it got
> > ignored.
> > Adjusted to log just port and eax (ecx was NOT tested for).
>
> I was actually meaning to apply this (at least until patch 6; I'm
> not convinced patch 7 makes sense to go in at this point), but
> noticed that comments by Konrad were still un-addressed. I'm
> also expecting some re-basing to be needed on top of Paul's
> emulation re-work.
>
> Wei, considering that this had already missed 4.5, would you
> be okay to take at least first 6 patches (which seem reasonably
> risk free) after the freeze?
>
Yes. I think it is fine, but please do it as soon as possible,
preferably within next week.
Wei.
> Jan
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 2/8] xen: Add support for VMware cpuid leaves
2015-07-07 15:03 ` Konrad Rzeszutek Wilk
@ 2015-07-28 2:54 ` Don Slutz
0 siblings, 0 replies; 29+ messages in thread
From: Don Slutz @ 2015-07-28 2:54 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On 07/07/15 11:03, Konrad Rzeszutek Wilk wrote:
>>>> diff --git a/xen/arch/x86/hvm/vmware/cpuid.c b/xen/arch/x86/hvm/vmware/cpuid.c
>>>> new file mode 100644
>>>> index 0000000..0dff36b
>>>> --- /dev/null
>>>> +++ b/xen/arch/x86/hvm/vmware/cpuid.c
>>>> @@ -0,0 +1,77 @@
>>>> +/*
>>>> + * arch/x86/hvm/vmware/cpuid.c
>>>> + *
>>>> + * Copyright (C) 2012 Verizon Corporation
>>>
>>> s/2012/2015/ ?
>>
>> The code here was written in 2012, 1st posted on xen-devel on Thu, 12
>> Dec 2013 14:15:11 -0500
>> ([Xen-devel] [RFC PATCH 03/10] Add cpuid_vmware_leaves), and this same
>> code has been posted in 2014 and 2015.
>>
>> So either no change or 2012-2015. Which do you want?
>
> It is up to you. Usually one posts with the latest copyright date.
>
Since I will need to re-post series do to merge issues, I will change to
2012-2015.
-Don Slutz
P.S. Sorry about the slow response, was on vacation.
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 3/8] tools: Add vmware_hwver support
2015-07-07 15:59 ` Ian Campbell
@ 2015-07-28 2:58 ` Don Slutz
0 siblings, 0 replies; 29+ messages in thread
From: Don Slutz @ 2015-07-28 2:58 UTC (permalink / raw)
To: Ian Campbell, Konrad Rzeszutek Wilk
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Jun Nakajima,
Stefano Stabellini, George Dunlap, Ian Jackson, Eddie Dong,
Don Slutz, xen-devel, Aravind Gopalakrishnan, Jan Beulich,
Andrew Cooper, Boris Ostrovsky, Suravee Suthikulpanit
On 07/07/15 11:59, Ian Campbell wrote:
> On Wed, 2015-07-01 at 16:14 -0400, Konrad Rzeszutek Wilk wrote:
>> On Sat, Jun 27, 2015 at 07:27:40PM -0400, Don Slutz wrote:
>>> From: Don Slutz <dslutz@verizon.com>
>>> +In a .ovf it is part of the value of vssd:VirtualSystemType.
>>> +For vssd:VirtualSystemType == vmx-07, vmware_hwver = 7.
>>> +
>>> +=back
>>> +
>>
>> Perhaps add 'Recommended value is 7' ?
>
> Assuming that is a good recommendation then that might be a good
> enhancement for a future patch. Or perhaps some sort of simple mapping
> from VMware product versions/names to a suitable value to use, unless
> that means having to think about trademarks etc.
>
I would be happy to continue this as a new patch.
-Don Slutz
> Ian.
>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT
2015-07-07 15:02 ` Konrad Rzeszutek Wilk
@ 2015-07-28 2:59 ` Don Slutz
0 siblings, 0 replies; 29+ messages in thread
From: Don Slutz @ 2015-07-28 2:59 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk
Cc: Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, Jun Nakajima, Eddie Dong, Ian Jackson,
Don Slutz, xen-devel, George Dunlap, Aravind Gopalakrishnan,
Jan Beulich, Andrew Cooper, Boris Ostrovsky,
Suravee Suthikulpanit
On 07/07/15 11:02, Konrad Rzeszutek Wilk wrote:
>>>> diff --git a/xen/include/public/hvm/params.h b/xen/include/public/hvm/params.h
>>>> index 7c73089..130eba9 100644
>>>> --- a/xen/include/public/hvm/params.h
>>>> +++ b/xen/include/public/hvm/params.h
>>>> @@ -50,6 +50,8 @@
>>>> #define HVM_PARAM_PAE_ENABLED 4
>>>> #define HVM_PARAM_IOREQ_PFN 5
>>>> +/* Extra vmport PFN. */
>>> s/Extra/optional/ ?
>>
>> I do not think "optional" is good to use here. This PFN is required for
>> QEMU (or any other ioreq server) to handle vmport access. If you have a
>> better name then "extra", please let me know.
>
> How about just 'VMWare Port PFN' ?
Will switch to this.
-Don Slutz
>>
>> -Don Slutz
>>
>>>> +#define HVM_PARAM_VMPORT_REGS_PFN 35
>>>> #define HVM_PARAM_BUFIOREQ_PFN 6
>>>> #define HVM_PARAM_BUFIOREQ_EVTCHN 26
>>>> @@ -187,6 +189,6 @@
>>>> /* Location of the VM Generation ID in guest physical address space. */
>>>> #define HVM_PARAM_VM_GENERATION_ID_ADDR 34
>>>> -#define HVM_NR_PARAMS 35
>>>> +#define HVM_NR_PARAMS 36
>>>> #endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */
>>>> --
>>>> 1.8.3.1
>>>>
>>
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 0/8] Xen VMware tools support
2015-07-10 9:07 ` Wei Liu
@ 2015-07-28 3:18 ` Don Slutz
2015-07-28 14:32 ` Konrad Rzeszutek Wilk
0 siblings, 1 reply; 29+ messages in thread
From: Don Slutz @ 2015-07-28 3:18 UTC (permalink / raw)
To: Wei Liu, Jan Beulich
Cc: Jun Nakajima, Tim Deegan, Kevin Tian, Keir Fraser, Ian Campbell,
Stefano Stabellini, George Dunlap, Andrew Cooper, Ian Jackson,
xen-devel, Eddie Dong, Aravind Gopalakrishnan,
Suravee Suthikulpanit, Boris Ostrovsky
On 07/10/15 05:07, Wei Liu wrote:
> On Fri, Jul 10, 2015 at 09:40:59AM +0100, Jan Beulich wrote:
>>>>> On 28.06.15 at 01:27, <don.slutz@gmail.com> wrote:
>>> Changes v11 to v12:
>>> Rebased on staging.
>>>
>>
>> I was actually meaning to apply this (at least until patch 6; I'm
>> not convinced patch 7 makes sense to go in at this point), but
>> noticed that comments by Konrad were still un-addressed. I'm
>> also expecting some re-basing to be needed on top of Paul's
>> emulation re-work.
>>
My understanding of Konrad's comments are:
For patch #2:
adjust a comment that is removed in patch #4.
I did this, but failed to get it posted before leaving on
vacation (sigh).
Adjust copyright date from 2012 to 2015 (but not required).
For patch #3:
Perhaps add 'Recommended value is 7' ?
I did not take this as a required change.
For patch #7 (which also need a lot of merge work) I was waiting till
Paul's changes had been committed. Now that they are I will be working
on the merge,
>> Wei, considering that this had already missed 4.5, would you
>> be okay to take at least first 6 patches (which seem reasonably
>> risk free) after the freeze?
>>
>
> Yes. I think it is fine, but please do it as soon as possible,
> preferably within next week.
>
I see that this did not happen. Do you want me to re-post just 1 to 6
that I have now rebased? (the improvement to the doc, Ian said could be
a follow on patch)
-Don Slutz
> Wei.
>
>> Jan
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v12 0/8] Xen VMware tools support
2015-07-28 3:18 ` Don Slutz
@ 2015-07-28 14:32 ` Konrad Rzeszutek Wilk
0 siblings, 0 replies; 29+ messages in thread
From: Konrad Rzeszutek Wilk @ 2015-07-28 14:32 UTC (permalink / raw)
To: Don Slutz
Cc: Jun Nakajima, Tim Deegan, Kevin Tian, Wei Liu, Ian Campbell,
Stefano Stabellini, George Dunlap, Andrew Cooper, Ian Jackson,
xen-devel, Jan Beulich, Eddie Dong, Aravind Gopalakrishnan,
Suravee Suthikulpanit, Keir Fraser, Boris Ostrovsky
On Mon, Jul 27, 2015 at 11:18:03PM -0400, Don Slutz wrote:
> On 07/10/15 05:07, Wei Liu wrote:
> > On Fri, Jul 10, 2015 at 09:40:59AM +0100, Jan Beulich wrote:
> >>>>> On 28.06.15 at 01:27, <don.slutz@gmail.com> wrote:
> >>> Changes v11 to v12:
> >>> Rebased on staging.
> >>>
>
>
> >>
> >> I was actually meaning to apply this (at least until patch 6; I'm
> >> not convinced patch 7 makes sense to go in at this point), but
> >> noticed that comments by Konrad were still un-addressed. I'm
> >> also expecting some re-basing to be needed on top of Paul's
> >> emulation re-work.
> >>
>
> My understanding of Konrad's comments are:
>
> For patch #2:
> adjust a comment that is removed in patch #4.
> I did this, but failed to get it posted before leaving on
> vacation (sigh).
>
> Adjust copyright date from 2012 to 2015 (but not required).
>
>
> For patch #3:
> Perhaps add 'Recommended value is 7' ?
> I did not take this as a required change.
>
>
> For patch #7 (which also need a lot of merge work) I was waiting till
> Paul's changes had been committed. Now that they are I will be working
> on the merge,
>
> >> Wei, considering that this had already missed 4.5, would you
> >> be okay to take at least first 6 patches (which seem reasonably
> >> risk free) after the freeze?
> >>
> >
> > Yes. I think it is fine, but please do it as soon as possible,
> > preferably within next week.
> >
>
> I see that this did not happen. Do you want me to re-post just 1 to 6
> that I have now rebased? (the improvement to the doc, Ian said could be
> a follow on patch)
Welcome back from vacation!
I think Wei was hoping to get your patches before the feature freeze.
Now that ship has sailed ...
>
> -Don Slutz
>
> > Wei.
> >
> >> Jan
^ permalink raw reply [flat|nested] 29+ messages in thread
end of thread, other threads:[~2015-07-28 14:32 UTC | newest]
Thread overview: 29+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-06-27 23:27 [PATCH v12 0/8] Xen VMware tools support Don Slutz
2015-06-27 23:27 ` [PATCH v12 1/8] tools: Add vga=vmware Don Slutz
2015-07-07 15:54 ` Ian Campbell
2015-06-27 23:27 ` [PATCH v12 2/8] xen: Add support for VMware cpuid leaves Don Slutz
2015-07-01 20:01 ` Konrad Rzeszutek Wilk
2015-07-03 22:03 ` Don Slutz
2015-07-07 15:03 ` Konrad Rzeszutek Wilk
2015-07-28 2:54 ` Don Slutz
2015-07-07 15:56 ` Ian Campbell
2015-06-27 23:27 ` [PATCH v12 3/8] tools: Add vmware_hwver support Don Slutz
2015-07-01 20:14 ` Konrad Rzeszutek Wilk
2015-07-07 15:59 ` Ian Campbell
2015-07-28 2:58 ` Don Slutz
2015-06-27 23:27 ` [PATCH v12 4/8] vmware: Add VMware provided include file Don Slutz
2015-06-27 23:27 ` [PATCH v12 5/8] xen: Add vmware_port support Don Slutz
2015-06-27 23:27 ` [PATCH v12 6/8] tools: " Don Slutz
2015-07-07 16:01 ` Ian Campbell
2015-06-27 23:27 ` [PATCH v12 7/8] Add IOREQ_TYPE_VMWARE_PORT Don Slutz
2015-07-01 20:49 ` Konrad Rzeszutek Wilk
2015-07-03 14:55 ` Don Slutz
2015-07-07 15:02 ` Konrad Rzeszutek Wilk
2015-07-28 2:59 ` Don Slutz
2015-06-27 23:27 ` [PATCH v12 8/8] Add xentrace to vmware_port Don Slutz
2015-06-29 14:54 ` Andrew Cooper
2015-06-29 16:11 ` Don Slutz
2015-07-10 8:40 ` [PATCH v12 0/8] Xen VMware tools support Jan Beulich
2015-07-10 9:07 ` Wei Liu
2015-07-28 3:18 ` Don Slutz
2015-07-28 14:32 ` Konrad Rzeszutek Wilk
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.