[PATCH v4 0/3] use wincap in format_proc_cpuinfo for user_shstk

[PATCH v4 0/3] use wincap in format_proc_cpuinfo for user_shstk

[Brian Inglis]

In test for for AMD/Intel Control flow Enforcement Technology user mode
shadow stack support replace Windows version tests with test of wincap
member addition has_user_shstk with Windows version dependent value

Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>

Brian Inglis (3):
  wincap.h: add wincap member has_user_shstk
  wincap.cc: set wincap member has_user_shstk true for 2004+
  fhandler/proc.cc: use wincap.has_user_shstk

 winsup/cygwin/fhandler/proc.cc        |  8 ++++----
 winsup/cygwin/local_includes/wincap.h |  2 ++
 winsup/cygwin/wincap.cc               | 10 ++++++++++
 3 files changed, 16 insertions(+), 4 deletions(-)

-- 
2.39.0

[PATCH v4 1/3] wincap.h: add wincap member has_user_shstk

[Brian Inglis]

Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
---
 winsup/cygwin/local_includes/wincap.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/winsup/cygwin/local_includes/wincap.h b/winsup/cygwin/local_includes/wincap.h
index 29a7a63de7f6..c14872787ca2 100644
--- a/winsup/cygwin/local_includes/wincap.h
+++ b/winsup/cygwin/local_includes/wincap.h
@@ -32,6 +32,7 @@ struct wincaps
     unsigned has_linux_tcp_keepalive_sockopts			: 1;
     unsigned has_tcp_maxrtms					: 1;
     unsigned has_con_broken_tabs				: 1;
+    unsigned has_user_shstk					: 1;
   };
 };
 
@@ -84,6 +85,7 @@ public:
   bool	IMPLEMENT (has_linux_tcp_keepalive_sockopts)
   bool	IMPLEMENT (has_tcp_maxrtms)
   bool	IMPLEMENT (has_con_broken_tabs)
+  bool	IMPLEMENT (has_user_shstk)
 
   void disable_case_sensitive_dirs ()
   {
-- 
2.39.0

[PATCH v4 2/3] wincap.cc: set wincap member has_user_shstk true for 2004+

[Brian Inglis]

Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
---
 winsup/cygwin/wincap.cc | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/winsup/cygwin/wincap.cc b/winsup/cygwin/wincap.cc
index 91d5d9df8889..30d9c14e8d3b 100644
--- a/winsup/cygwin/wincap.cc
+++ b/winsup/cygwin/wincap.cc
@@ -31,6 +31,7 @@ static const wincaps wincap_8_1 = {
     has_linux_tcp_keepalive_sockopts:false,
     has_tcp_maxrtms:false,
     has_con_broken_tabs:false,
+    has_user_shstk:false,
   },
 };
 
@@ -52,6 +53,7 @@ static const wincaps  wincap_10_1507 = {
     has_linux_tcp_keepalive_sockopts:false,
     has_tcp_maxrtms:false,
     has_con_broken_tabs:false,
+    has_user_shstk:false,
   },
 };
 
@@ -73,6 +75,7 @@ static const wincaps  wincap_10_1607 = {
     has_linux_tcp_keepalive_sockopts:false,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:false,
+    has_user_shstk:false,
   },
 };
 
@@ -94,6 +97,7 @@ static const wincaps wincap_10_1703 = {
     has_linux_tcp_keepalive_sockopts:false,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:true,
+    has_user_shstk:false,
   },
 };
 
@@ -115,6 +119,7 @@ static const wincaps wincap_10_1709 = {
     has_linux_tcp_keepalive_sockopts:true,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:true,
+    has_user_shstk:false,
   },
 };
 
@@ -136,6 +141,7 @@ static const wincaps wincap_10_1803 = {
     has_linux_tcp_keepalive_sockopts:true,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:true,
+    has_user_shstk:false,
   },
 };
 
@@ -157,6 +163,7 @@ static const wincaps wincap_10_1809 = {
     has_linux_tcp_keepalive_sockopts:true,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:true,
+    has_user_shstk:false,
   },
 };
 
@@ -178,6 +185,7 @@ static const wincaps wincap_10_1903 = {
     has_linux_tcp_keepalive_sockopts:true,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:true,
+    has_user_shstk:false,
   },
 };
 
@@ -199,6 +207,7 @@ static const wincaps wincap_10_2004 = {
     has_linux_tcp_keepalive_sockopts:true,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:true,
+    has_user_shstk:true,
   },
 };
 
@@ -220,6 +229,7 @@ static const wincaps wincap_11 = {
     has_linux_tcp_keepalive_sockopts:true,
     has_tcp_maxrtms:true,
     has_con_broken_tabs:false,
+    has_user_shstk:true,
   },
 };
 
-- 
2.39.0

[PATCH v4 3/3] fhandler/proc.cc: use wincap.has_user_shstk

[Brian Inglis]

Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
---
 winsup/cygwin/fhandler/proc.cc | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/winsup/cygwin/fhandler/proc.cc b/winsup/cygwin/fhandler/proc.cc
index 3c79762e0fbd..cbc49a12a417 100644
--- a/winsup/cygwin/fhandler/proc.cc
+++ b/winsup/cygwin/fhandler/proc.cc
@@ -1486,12 +1486,12 @@ format_proc_cpuinfo (void *, char *&destbuf)
 
 /*	  ftcprint (features1,  6, "split_lock_detect");*//* MSR_TEST_CTRL split lock */
 
-      /* cpuid 0x00000007 ecx & Windows [20]20H1/[20]2004+ */
-      if (maxf >= 0x00000007 && wincap.osname () >= "10.0"
-					 && wincap.build_number () >= 19041)
+      /* Windows [20]20H1/[20]2004/19041 user shadow stack */
+      if (maxf >= 0x00000007 && wincap.has_user_shstk ())
         {
+	  /* cpuid 0x00000007 ecx CET shadow stack */
 	  cpuid (&unused, &unused, &features1, &unused, 0x00000007, 0);
-	  ftcprint (features1,  7, "user_shstk");	/* "user shadow stack" */
+	  ftcprint (features1,  7, "user_shstk");	/* user shadow stack */
 	}
 
       /* cpuid 0x00000007:1 eax */
-- 
2.39.0

Re: [PATCH v4 0/3] use wincap in format_proc_cpuinfo for user_shstk

[Corinna Vinschen]

On Jun 19 12:15, Brian Inglis wrote:
> In test for for AMD/Intel Control flow Enforcement Technology user mode
> shadow stack support replace Windows version tests with test of wincap
> member addition has_user_shstk with Windows version dependent value
> 
> Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
> Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
> 
> Brian Inglis (3):
>   wincap.h: add wincap member has_user_shstk
>   wincap.cc: set wincap member has_user_shstk true for 2004+
>   fhandler/proc.cc: use wincap.has_user_shstk
> 
>  winsup/cygwin/fhandler/proc.cc        |  8 ++++----
>  winsup/cygwin/local_includes/wincap.h |  2 ++
>  winsup/cygwin/wincap.cc               | 10 ++++++++++
>  3 files changed, 16 insertions(+), 4 deletions(-)

Never mind, I fixed the remaining problems.  Thanks for the patch,
I pushed it with slight modifications to the commit messages.

I'm a bit puzzled if my original mail
https://cygwin.com/pipermail/cygwin-patches/2023q2/012280.html
was really that unclear.  Reiterating for the records:

- Commit messages should really try to explain why the patch is made and
  what it's good for. In case of fixing a bug, the bug should be explained
  and, ideally, explain why the patch is the better solution.

- If a patch fixes an older bug, it should say so and point out the
  commit which introduced the bug using the Fixes: tag.  The format
  is
  
    Fixes: <12-digit-SHA1> ("<commit headline>")

  The parens and quoting chars are part of the format.

- Every patch should contain a Signed-off-by: to indicate that
  you did the patch by yourself.  That's easily automated by
  using `git commit -s'.

- Other Tags like "Reported-by:" or "Tested-by:" are nice to have,
  but not essential.


Thanks,
Corinna

Re: [PATCH v4 0/3] use wincap in format_proc_cpuinfo for user_shstk

[Brian Inglis]

On 2023-06-20 02:22, Corinna Vinschen wrote:
> On Jun 19 12:15, Brian Inglis wrote:
>> In test for for AMD/Intel Control flow Enforcement Technology user mode
>> shadow stack support replace Windows version tests with test of wincap
>> member addition has_user_shstk with Windows version dependent value
>>
>> Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
>> Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
>>
>> Brian Inglis (3):
>>    wincap.h: add wincap member has_user_shstk
>>    wincap.cc: set wincap member has_user_shstk true for 2004+
>>    fhandler/proc.cc: use wincap.has_user_shstk
>>
>>   winsup/cygwin/fhandler/proc.cc        |  8 ++++----
>>   winsup/cygwin/local_includes/wincap.h |  2 ++
>>   winsup/cygwin/wincap.cc               | 10 ++++++++++
>>   3 files changed, 16 insertions(+), 4 deletions(-)
> 
> Never mind, I fixed the remaining problems.  Thanks for the patch,
> I pushed it with slight modifications to the commit messages.
> 
> I'm a bit puzzled if my original mail
> https://cygwin.com/pipermail/cygwin-patches/2023q2/012280.html
> was really that unclear.  Reiterating for the records:
> 
> - Commit messages should really try to explain why the patch is made and
>    what it's good for. In case of fixing a bug, the bug should be explained
>    and, ideally, explain why the patch is the better solution.
> 
> - If a patch fixes an older bug, it should say so and point out the
>    commit which introduced the bug using the Fixes: tag.  The format
>    is
>    
>      Fixes: <12-digit-SHA1> ("<commit headline>")
> 
>    The parens and quoting chars are part of the format.
> 
> - Every patch should contain a Signed-off-by: to indicate that
>    you did the patch by yourself.  That's easily automated by
>    using `git commit -s'.
> 
> - Other Tags like "Reported-by:" or "Tested-by:" are nice to have,
>    but not essential.

>> - For obvious reasons, the message text in your cover message won't make
>>   it into the git repo.  However, the commit messages in git should
>>   reflect why the change was made, so a future interested reader has
>>   a chance to understand why a change was made.

Not obvious to me unfortunately!

>> - As I already mentioned a couple of times on this list (but not
>>   lately), it would be great if you could always add a "Signed-off-by:"
>>   line.

I added that to my config.

>> - Also, given this patch fixes an earlier patch, it should contain
>>   a line
>> 
>>     Fixes: <12-digit-SHA1> ("commit headline")

ASSUME NOTHING (as we used to write in masm) about me, git, background info, or 
conventions: man page or link refs please?

I personally used CVS then Hg, and needed BK for another project.

I only use git seriously for this project, as most businesses supported at best 
CVS, most projects relied on backups of network shares, and saw little value in 
source control, even in software cos; others were anal about it, requiring 
diffs, but did not trust developers (maybe just contractors) with access to 
their secure vaults!
[It made for interesting chats, some after I was gone, when their test builds 
were still showing bugs I had fixed, tested, and submitted source changes for 
weeks earlier!]

After trying to earlier track down useful docs I just managed to find:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst

which presumably is the background I need?

Some stuff is also mentioned in:

https://git.kernel.org/pub/scm/git/git.git/tree/Documentation/SubmittingPatches

which may have a more explanatory and practical focus.

-- 
Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada

La perfection est atteinte                   Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter  not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer     but when there is no more to cut
                                 -- Antoine de Saint-Exupéry

Re: [PATCH v4 0/3] use wincap in format_proc_cpuinfo for user_shstk

[Corinna Vinschen]

On Jun 20 22:50, Brian Inglis wrote:
> On 2023-06-20 02:22, Corinna Vinschen wrote:
> > On Jun 19 12:15, Brian Inglis wrote:
> > > In test for for AMD/Intel Control flow Enforcement Technology user mode
> > > shadow stack support replace Windows version tests with test of wincap
> > > member addition has_user_shstk with Windows version dependent value
> > > 
> > > Fixes: 41fdb869f998 fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
> > > Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
> > > 
> > > Brian Inglis (3):
> > >    wincap.h: add wincap member has_user_shstk
> > >    wincap.cc: set wincap member has_user_shstk true for 2004+
> > >    fhandler/proc.cc: use wincap.has_user_shstk
> > > 
> > >   winsup/cygwin/fhandler/proc.cc        |  8 ++++----
> > >   winsup/cygwin/local_includes/wincap.h |  2 ++
> > >   winsup/cygwin/wincap.cc               | 10 ++++++++++
> > >   3 files changed, 16 insertions(+), 4 deletions(-)
> > 
> > Never mind, I fixed the remaining problems.  Thanks for the patch,
> > I pushed it with slight modifications to the commit messages.
> > 
> > I'm a bit puzzled if my original mail
> > https://cygwin.com/pipermail/cygwin-patches/2023q2/012280.html
> > was really that unclear.  Reiterating for the records:
> > 
> > - Commit messages should really try to explain why the patch is made and
> >    what it's good for. In case of fixing a bug, the bug should be explained
> >    and, ideally, explain why the patch is the better solution.
> > 
> > - If a patch fixes an older bug, it should say so and point out the
> >    commit which introduced the bug using the Fixes: tag.  The format
> >    is
> >      Fixes: <12-digit-SHA1> ("<commit headline>")
> > 
> >    The parens and quoting chars are part of the format.
> > 
> > - Every patch should contain a Signed-off-by: to indicate that
> >    you did the patch by yourself.  That's easily automated by
> >    using `git commit -s'.
> > 
> > - Other Tags like "Reported-by:" or "Tested-by:" are nice to have,
> >    but not essential.
> 
> > > - For obvious reasons, the message text in your cover message won't make
> > >   it into the git repo.  However, the commit messages in git should
> > >   reflect why the change was made, so a future interested reader has
> > >   a chance to understand why a change was made.
> 
> Not obvious to me unfortunately!

No worries, we're all learning while we're going along.

> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst

This is pretty well written, and it very nicely explains how to write
commit messages and use tagging.

Writing useful commit messages helps other people a lot to understand
why a patch was made, especially in tricky error cases (locking issues
are hard to follow, a good explanation is key).

If you ask me, there's no such thing as a too long commit message.

We're a small group so we're not supposed to follow the above document
to the core.  However, things like "Fixes:" are really great, because
they connect a patch with a history of other patches and allow an aha
effect when checking on the referenced older patch.  I can honestely
tell you that it already helped me a lot when working on the Linux
kernel.  That's why I'd like to make this standard for our small project
here, too.

And one of the core expressions used in this doc is this:

  Don't get discouraged - or impatient


:)
Corinna

Re: [PATCH v4 0/3] use wincap in format_proc_cpuinfo for user_shstk

[Brian Inglis]

On 2023-06-21 02:22, Corinna Vinschen wrote:
> On Jun 20 22:50, Brian Inglis wrote:
>> On 2023-06-20 02:22, Corinna Vinschen wrote:
>>> Never mind, I fixed the remaining problems.  Thanks for the patch,
>>> I pushed it with slight modifications to the commit messages.
>>>
>>> I'm a bit puzzled if my original mail
>>> https://cygwin.com/pipermail/cygwin-patches/2023q2/012280.html
>>> was really that unclear.  Reiterating for the records:
>>>
>>> - Commit messages should really try to explain why the patch is made and
>>>     what it's good for. In case of fixing a bug, the bug should be explained
>>>     and, ideally, explain why the patch is the better solution.
>>>
>>> - If a patch fixes an older bug, it should say so and point out the
>>>     commit which introduced the bug using the Fixes: tag.  The format
>>>     is
>>>       Fixes: <12-digit-SHA1> ("<commit headline>")
>>>
>>>     The parens and quoting chars are part of the format.
>>>
>>> - Every patch should contain a Signed-off-by: to indicate that
>>>     you did the patch by yourself.  That's easily automated by
>>>     using `git commit -s'.
>>>
>>> - Other Tags like "Reported-by:" or "Tested-by:" are nice to have,
>>>     but not essential.
>>
>>>> - For obvious reasons, the message text in your cover message won't make
>>>>    it into the git repo.  However, the commit messages in git should
>>>>    reflect why the change was made, so a future interested reader has
>>>>    a chance to understand why a change was made.
>>
>> Not obvious to me unfortunately!
> 
> No worries, we're all learning while we're going along.
> 
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst
> 
> This is pretty well written, and it very nicely explains how to write
> commit messages and use tagging.
> 
> Writing useful commit messages helps other people a lot to understand
> why a patch was made, especially in tricky error cases (locking issues
> are hard to follow, a good explanation is key).
> 
> If you ask me, there's no such thing as a too long commit message.
> 
> We're a small group so we're not supposed to follow the above document
> to the core.  However, things like "Fixes:" are really great, because
> they connect a patch with a history of other patches and allow an aha
> effect when checking on the referenced older patch.  I can honestely
> tell you that it already helped me a lot when working on the Linux
> kernel.  That's why I'd like to make this standard for our small project
> here, too.
> 
> And one of the core expressions used in this doc is this:
>    Don't get discouraged - or impatient
> :)

Cheers - never discouraged - a bit impatient because often frustrated - 
sometimes confused - like Alice and the White Knight's song "Haddocks' Eyes" - 
finding that the /footers/ are sometimes called /headers/ and other times /tags/ 
but actually named *trailers* finally gave more useful results where /fixes/ did 
not! ;^>

-- 
Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada

La perfection est atteinte                   Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter  not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer     but when there is no more to cut
                                 -- Antoine de Saint-Exupéry