diff mbox series

NEWS: Intel CET support hasn't be validated for i686

Message ID CAMe9rOpBM0-tanrxEu9Vk=jdHpg=GGugP3TSm9BD+KUfDCVqEQ@mail.gmail.com
State New
Headers show
Series NEWS: Intel CET support hasn't be validated for i686 | expand

Commit Message

H.J. Lu July 19, 2018, 5:12 p.m. UTC
On Wed, Jul 18, 2018 at 2:16 PM, Joseph Myers <joseph@codesourcery.com> wrote:
> On Wed, 18 Jul 2018, H.J. Lu wrote:
>
>> 2. On 64-bit CET SDV under CET kernel with --enable-cet
>>      a. Only x86_64 and x32 tested.
>>      b. i686 isn't tested since 64-bit CET SDV doesn't support i686.
>
> So the key gap in test coverage is the lack of any evidence that the i686
> code does actually work on CET processors.  I think that at least merits a
> warning in the NEWS and install.texi entries about the unvalidated nature
> of the i686 CET support (which would be removed from install.texi at a
> later date once the i686 support has been validated, on a simulator or
> hardware, and any issues resolved).
>

How about this?

Comments

Carlos O'Donell July 19, 2018, 5:14 p.m. UTC | #1
On 07/19/2018 01:12 PM, H.J. Lu wrote:
> On Wed, Jul 18, 2018 at 2:16 PM, Joseph Myers <joseph@codesourcery.com> wrote:
>> On Wed, 18 Jul 2018, H.J. Lu wrote:
>>
>>> 2. On 64-bit CET SDV under CET kernel with --enable-cet
>>>      a. Only x86_64 and x32 tested.
>>>      b. i686 isn't tested since 64-bit CET SDV doesn't support i686.
>> So the key gap in test coverage is the lack of any evidence that the i686
>> code does actually work on CET processors.  I think that at least merits a
>> warning in the NEWS and install.texi entries about the unvalidated nature
>> of the i686 CET support (which would be removed from install.texi at a
>> later date once the i686 support has been validated, on a simulator or
>> hardware, and any issues resolved).
>>
> How about this?

This looks good to me and is very clear about the i686 support.

OK for 2.28.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>
 
> -- H.J.
> 
> 
> 0001-NEWS-Intel-CET-support-hasn-t-be-validated-for-i686.patch
> 
> 
> From 3191cbcf1ff664aa2f12a6d83f1349639e051208 Mon Sep 17 00:00:00 2001
> From: "H.J. Lu" <hjl.tools@gmail.com>
> Date: Wed, 18 Jul 2018 14:55:46 -0700
> Subject: [PATCH] NEWS: Intel CET support hasn't be validated for i686
> 
> ---
>  NEWS | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/NEWS b/NEWS
> index daef815ae7..b4055bf99e 100644
> --- a/NEWS
> +++ b/NEWS
> @@ -17,7 +17,10 @@ Major new features:
>    feature is currently supported on i386, x86_64 and x32 with GCC 8 and
>    binutils 2.29 or later.  Note that CET-enabled glibc requires CPUs
>    capable of multi-byte NOPs, like x86-64 processors as well as Intel
> -  Pentium Pro or newer.
> +  Pentium Pro or newer.  NOTE: --enable-cet has been tested for i686,
> +  x86_64 and x32 on non-CET processors.  --enable-cet has been tested
> +  for x86_64 and x32 on CET SDVs.  But Intel CET support hasn't be
> +  validated for i686.
>  
>  * The GNU C Library now has correct support for ABSOLUTE symbols
>    (SHN_ABS-relative symbols).  Previously such ABSOLUTE symbols were
> -- 2.17.1
Joseph Myers July 19, 2018, 5:15 p.m. UTC | #2
On Thu, 19 Jul 2018, H.J. Lu wrote:

> On Wed, Jul 18, 2018 at 2:16 PM, Joseph Myers <joseph@codesourcery.com> wrote:
> > On Wed, 18 Jul 2018, H.J. Lu wrote:
> >
> >> 2. On 64-bit CET SDV under CET kernel with --enable-cet
> >>      a. Only x86_64 and x32 tested.
> >>      b. i686 isn't tested since 64-bit CET SDV doesn't support i686.
> >
> > So the key gap in test coverage is the lack of any evidence that the i686
> > code does actually work on CET processors.  I think that at least merits a
> > warning in the NEWS and install.texi entries about the unvalidated nature
> > of the i686 CET support (which would be removed from install.texi at a
> > later date once the i686 support has been validated, on a simulator or
> > hardware, and any issues resolved).
> >
> 
> How about this?

You're missing the update to install.texi.

People reading NEWS for this release need to see the caveat about the new 
feature described.

People reading INSTALL for this or a later release need to see the caveat 
about the configure option they are reading the documentation for, for as 
long as that caveat applies.

Both places need the warning.
Dmitry V. Levin July 19, 2018, 5:19 p.m. UTC | #3
On Thu, Jul 19, 2018 at 10:12:09AM -0700, H.J. Lu wrote:
[...]
> diff --git a/NEWS b/NEWS
> index daef815ae7..b4055bf99e 100644
> --- a/NEWS
> +++ b/NEWS
> @@ -17,7 +17,10 @@ Major new features:
>    feature is currently supported on i386, x86_64 and x32 with GCC 8 and
>    binutils 2.29 or later.  Note that CET-enabled glibc requires CPUs
>    capable of multi-byte NOPs, like x86-64 processors as well as Intel
> -  Pentium Pro or newer.
> +  Pentium Pro or newer.  NOTE: --enable-cet has been tested for i686,
> +  x86_64 and x32 on non-CET processors.  --enable-cet has been tested
> +  for x86_64 and x32 on CET SDVs.  But Intel CET support hasn't be

hasn't been?
H.J. Lu July 19, 2018, 5:51 p.m. UTC | #4
On Thu, Jul 19, 2018 at 10:19 AM, Dmitry V. Levin <ldv@altlinux.org> wrote:
> On Thu, Jul 19, 2018 at 10:12:09AM -0700, H.J. Lu wrote:
> [...]
>> diff --git a/NEWS b/NEWS
>> index daef815ae7..b4055bf99e 100644
>> --- a/NEWS
>> +++ b/NEWS
>> @@ -17,7 +17,10 @@ Major new features:
>>    feature is currently supported on i386, x86_64 and x32 with GCC 8 and
>>    binutils 2.29 or later.  Note that CET-enabled glibc requires CPUs
>>    capable of multi-byte NOPs, like x86-64 processors as well as Intel
>> -  Pentium Pro or newer.
>> +  Pentium Pro or newer.  NOTE: --enable-cet has been tested for i686,
>> +  x86_64 and x32 on non-CET processors.  --enable-cet has been tested
>> +  for x86_64 and x32 on CET SDVs.  But Intel CET support hasn't be
>
> hasn't been?
>
>

I am going to check in this updated patch.
Rical Jasan July 19, 2018, 5:56 p.m. UTC | #5
On 07/19/2018 10:51 AM, H.J. Lu wrote:
...
> diff --git a/NEWS b/NEWS
> index daef815ae7..c6541f2004 100644
> --- a/NEWS
> +++ b/NEWS
> @@ -17,7 +17,10 @@ Major new features:
>    feature is currently supported on i386, x86_64 and x32 with GCC 8 and
>    binutils 2.29 or later.  Note that CET-enabled glibc requires CPUs
>    capable of multi-byte NOPs, like x86-64 processors as well as Intel
> -  Pentium Pro or newer.
> +  Pentium Pro or newer.  NOTE: --enable-cet has been tested for i686,
> +  x86_64 and x32 on non-CET processors.  --enable-cet has been tested
> +  for x86_64 and x32 on CET SDVs.  But Intel CET support hasn't been

"SDVs, but"

> +  validated for i686.
>  
>  * The GNU C Library now has correct support for ABSOLUTE symbols
>    (SHN_ABS-relative symbols).  Previously such ABSOLUTE symbols were
> diff --git a/manual/install.texi b/manual/install.texi
> index 3a87ac8bb5..61d3b1324a 100644
> --- a/manual/install.texi
> +++ b/manual/install.texi
> @@ -148,6 +148,11 @@ Note that when CET is enabled, @theglibc{} requires CPUs capable of
>  multi-byte NOPs, like x86-64 processors as well as Intel Pentium Pro or
>  newer.
>  
> +NOTE: @option{--enable-cet} has been tested for i686, x86_64 and x32
> +on non-CET processors.  @option{--enable-cet} has been tested for
> +x86_64 and x32 on CET SDVs.  But Intel CET support hasn't been validated

Same here.

> +for i686.
> +
>  @item --disable-profile
>  Don't build libraries with profiling information.  You may want to use
>  this option if you don't plan to do profiling.
> -- 2.17.1
>
H.J. Lu July 19, 2018, 6:22 p.m. UTC | #6
On Thu, Jul 19, 2018 at 10:56 AM, Rical Jasan <rj@2c3t.io> wrote:
> On 07/19/2018 10:51 AM, H.J. Lu wrote:
> ...
>> diff --git a/NEWS b/NEWS
>> index daef815ae7..c6541f2004 100644
>> --- a/NEWS
>> +++ b/NEWS
>> @@ -17,7 +17,10 @@ Major new features:
>>    feature is currently supported on i386, x86_64 and x32 with GCC 8 and
>>    binutils 2.29 or later.  Note that CET-enabled glibc requires CPUs
>>    capable of multi-byte NOPs, like x86-64 processors as well as Intel
>> -  Pentium Pro or newer.
>> +  Pentium Pro or newer.  NOTE: --enable-cet has been tested for i686,
>> +  x86_64 and x32 on non-CET processors.  --enable-cet has been tested
>> +  for x86_64 and x32 on CET SDVs.  But Intel CET support hasn't been
>
> "SDVs, but"
>
>> +  validated for i686.
>>
>>  * The GNU C Library now has correct support for ABSOLUTE symbols
>>    (SHN_ABS-relative symbols).  Previously such ABSOLUTE symbols were
>> diff --git a/manual/install.texi b/manual/install.texi
>> index 3a87ac8bb5..61d3b1324a 100644
>> --- a/manual/install.texi
>> +++ b/manual/install.texi
>> @@ -148,6 +148,11 @@ Note that when CET is enabled, @theglibc{} requires CPUs capable of
>>  multi-byte NOPs, like x86-64 processors as well as Intel Pentium Pro or
>>  newer.
>>
>> +NOTE: @option{--enable-cet} has been tested for i686, x86_64 and x32
>> +on non-CET processors.  @option{--enable-cet} has been tested for
>> +x86_64 and x32 on CET SDVs.  But Intel CET support hasn't been validated
>
> Same here.
>

Like this?
Rical Jasan July 19, 2018, 6:23 p.m. UTC | #7
On 07/19/2018 11:22 AM, H.J. Lu wrote:
> Like this?

Yep.

Rical
diff mbox series

Patch

From 3191cbcf1ff664aa2f12a6d83f1349639e051208 Mon Sep 17 00:00:00 2001
From: "H.J. Lu" <hjl.tools@gmail.com>
Date: Wed, 18 Jul 2018 14:55:46 -0700
Subject: [PATCH] NEWS: Intel CET support hasn't be validated for i686

---
 NEWS | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/NEWS b/NEWS
index daef815ae7..b4055bf99e 100644
--- a/NEWS
+++ b/NEWS
@@ -17,7 +17,10 @@  Major new features:
   feature is currently supported on i386, x86_64 and x32 with GCC 8 and
   binutils 2.29 or later.  Note that CET-enabled glibc requires CPUs
   capable of multi-byte NOPs, like x86-64 processors as well as Intel
-  Pentium Pro or newer.
+  Pentium Pro or newer.  NOTE: --enable-cet has been tested for i686,
+  x86_64 and x32 on non-CET processors.  --enable-cet has been tested
+  for x86_64 and x32 on CET SDVs.  But Intel CET support hasn't be
+  validated for i686.
 
 * The GNU C Library now has correct support for ABSOLUTE symbols
   (SHN_ABS-relative symbols).  Previously such ABSOLUTE symbols were
-- 
2.17.1