[edk2-devel] Please Consider This PR for edk2-stable202505

[edk2-devel] Please Consider This PR for edk2-stable202505

[Oliver Smith-Denny via groups.io]

Hi Liming and Mike,

This PR https://github.com/tianocore/edk2/pull/11086
addresses a crash for the Linux shim v16 on debug
builds. When DebugClearMemory is set, DXE Core will
currently attempt to write to all freed pages, even
if they are not writeable. This patch ensures that
DXE Core will change the memory attributes of
the free pages before writing to them, or will
not write to them if it fails to.

Without this, users of shim v16 will experience
failures when using builds of edk2 that have
DebugClearMemory set when loading certain
Linux distributions. Please consider merging
this for edk2-stable202505.

Thanks,
Oliver


-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121347): https://edk2.groups.io/g/devel/message/121347
Mute This Topic: https://groups.io/mt/113148129/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-

回复: [edk2-devel] Please Consider This PR for edk2-stable202505

[gaoliming via groups.io]

I see this PR is still under discussion. Does it finalize?

> -----邮件原件-----
> 发件人: devel@edk2.groups.io <devel@edk2.groups.io> 代表 Oliver
> Smith-Denny via groups.io
> 发送时间: 2025年5月17日 0:25
> 收件人: devel@edk2.groups.io; Liming Gao <gaoliming@byosoft.com.cn>;
> Michael Kinney <michael.d.kinney@intel.com>
> 主题: [edk2-devel] Please Consider This PR for edk2-stable202505
> 
> Hi Liming and Mike,
> 
> This PR https://github.com/tianocore/edk2/pull/11086
> addresses a crash for the Linux shim v16 on debug
> builds. When DebugClearMemory is set, DXE Core will
> currently attempt to write to all freed pages, even
> if they are not writeable. This patch ensures that
> DXE Core will change the memory attributes of
> the free pages before writing to them, or will
> not write to them if it fails to.
> 
> Without this, users of shim v16 will experience
> failures when using builds of edk2 that have
> DebugClearMemory set when loading certain
> Linux distributions. Please consider merging
> this for edk2-stable202505.
> 
> Thanks,
> Oliver
> 
> 
> 
> 





-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121350): https://edk2.groups.io/g/devel/message/121350
Mute This Topic: https://groups.io/mt/113203971/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [edk2-devel] Please Consider This PR for edk2-stable202505

[Michael D Kinney via groups.io]

Hi Liming,

+stewards

It is still being reviewed.  This is an important fix and needs to
go in as soon as it is ready, but I do not think it should be considered
for the stable tag.  With the level of changes being discussed I would
prefer it go in immediately after the stable tag and get lots of testing
before the August stable tag.

Mike


> -----Original Message-----
> From: gaoliming <gaoliming@byosoft.com.cn>
> Sent: Monday, May 19, 2025 6:06 PM
> To: devel@edk2.groups.io; osde@linux.microsoft.com; Kinney, Michael D
> <michael.d.kinney@intel.com>
> Subject: 回复: [edk2-devel] Please Consider This PR for edk2-stable202505
> 
> I see this PR is still under discussion. Does it finalize?
> 
> > -----邮件原件-----
> > 发件人: devel@edk2.groups.io <devel@edk2.groups.io> 代表 Oliver
> > Smith-Denny via groups.io
> > 发送时间: 2025年5月17日 0:25
> > 收件人: devel@edk2.groups.io; Liming Gao <gaoliming@byosoft.com.cn>;
> > Michael Kinney <michael.d.kinney@intel.com>
> > 主题: [edk2-devel] Please Consider This PR for edk2-stable202505
> >
> > Hi Liming and Mike,
> >
> > This PR https://github.com/tianocore/edk2/pull/11086
> > addresses a crash for the Linux shim v16 on debug
> > builds. When DebugClearMemory is set, DXE Core will
> > currently attempt to write to all freed pages, even
> > if they are not writeable. This patch ensures that
> > DXE Core will change the memory attributes of
> > the free pages before writing to them, or will
> > not write to them if it fails to.
> >
> > Without this, users of shim v16 will experience
> > failures when using builds of edk2 that have
> > DebugClearMemory set when loading certain
> > Linux distributions. Please consider merging
> > this for edk2-stable202505.
> >
> > Thanks,
> > Oliver
> >
> >
> > 
> >
> 
> 



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121362): https://edk2.groups.io/g/devel/message/121362
Mute This Topic: https://groups.io/mt/113215139/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [edk2-devel] Please Consider This PR for edk2-stable202505

[Oliver Smith-Denny via groups.io]

On 5/20/2025 10:20 AM, Michael D Kinney via groups.io wrote:
> Hi Liming,
> 
> +stewards
> 
> It is still being reviewed.  This is an important fix and needs to
> go in as soon as it is ready, but I do not think it should be considered
> for the stable tag.  With the level of changes being discussed I would
> prefer it go in immediately after the stable tag and get lots of testing
> before the August stable tag.
> 

I agree with Mike. This is a longstanding issue that was just exposed
recently, but the fix has evolved into a riskier change that needs a lot
of testing.

Released Linux shim versions with DebugClearMemory enabled will be
broken when loading certain Linux payloads without this change (and
there is a risk without DebugClearMemory enabled that pages would
get allocated to the next caller of AllocatePages set to RO and cause
unknown crashes). I defer to the stewards on the need to have this
stable tag contain a fix for this case or for it to get more testing
first (again, my preference).

Thanks,
Oliver



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121363): https://edk2.groups.io/g/devel/message/121363
Mute This Topic: https://groups.io/mt/113215139/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [edk2-devel] Please Consider This PR for edk2-stable202505

[Michael Kubacki via groups.io]

I agree with postponing it until after the May stable tag.

On 5/20/2025 1:30 PM, Oliver Smith-Denny via groups.io wrote:
> On 5/20/2025 10:20 AM, Michael D Kinney via groups.io wrote:
>> Hi Liming,
>>
>> +stewards
>>
>> It is still being reviewed.  This is an important fix and needs to
>> go in as soon as it is ready, but I do not think it should be considered
>> for the stable tag.  With the level of changes being discussed I would
>> prefer it go in immediately after the stable tag and get lots of testing
>> before the August stable tag.
>>
> 
> I agree with Mike. This is a longstanding issue that was just exposed
> recently, but the fix has evolved into a riskier change that needs a lot
> of testing.
> 
> Released Linux shim versions with DebugClearMemory enabled will be
> broken when loading certain Linux payloads without this change (and
> there is a risk without DebugClearMemory enabled that pages would
> get allocated to the next caller of AllocatePages set to RO and cause
> unknown crashes). I defer to the stewards on the need to have this
> stable tag contain a fix for this case or for it to get more testing
> first (again, my preference).
> 
> Thanks,
> Oliver
> 
> 
> 
> 
> 



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121364): https://edk2.groups.io/g/devel/message/121364
Mute This Topic: https://groups.io/mt/113215139/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-

回复: [edk2-devel] Please Consider This PR for edk2-stable202505

[gaoliming via groups.io]

OK. I will not merge it for this stable tag 202505.

> -----邮件原件-----
> 发件人: devel@edk2.groups.io <devel@edk2.groups.io> 代表 Michael
> Kubacki via groups.io
> 发送时间: 2025年5月21日 1:39
> 收件人: devel@edk2.groups.io; osde@linux.microsoft.com;
> michael.d.kinney@intel.com; gaoliming <gaoliming@byosoft.com.cn>; Leif
> Lindholm <leif.lindholm@oss.qualcomm.com>; Ard Biesheuvel
> <ardb@kernel.org>; Kubacki, Michael <michael.kubacki@microsoft.com>;
> Andrew Fish <afish@apple.com>
> 主题: Re: [edk2-devel] Please Consider This PR for edk2-stable202505
> 
> I agree with postponing it until after the May stable tag.
> 
> On 5/20/2025 1:30 PM, Oliver Smith-Denny via groups.io wrote:
> > On 5/20/2025 10:20 AM, Michael D Kinney via groups.io wrote:
> >> Hi Liming,
> >>
> >> +stewards
> >>
> >> It is still being reviewed.  This is an important fix and needs to
> >> go in as soon as it is ready, but I do not think it should be considered
> >> for the stable tag.  With the level of changes being discussed I would
> >> prefer it go in immediately after the stable tag and get lots of testing
> >> before the August stable tag.
> >>
> >
> > I agree with Mike. This is a longstanding issue that was just exposed
> > recently, but the fix has evolved into a riskier change that needs a lot
> > of testing.
> >
> > Released Linux shim versions with DebugClearMemory enabled will be
> > broken when loading certain Linux payloads without this change (and
> > there is a risk without DebugClearMemory enabled that pages would
> > get allocated to the next caller of AllocatePages set to RO and cause
> > unknown crashes). I defer to the stewards on the need to have this
> > stable tag contain a fix for this case or for it to get more testing
> > first (again, my preference).
> >
> > Thanks,
> > Oliver
> >
> >
> >
> >
> >
> 
> 
> 
> 
> 





-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121370): https://edk2.groups.io/g/devel/message/121370
Mute This Topic: https://groups.io/mt/113239781/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-