[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

This bug was fixed in the package linux-ti-omap - 2.6.33-500.4 --- linux-ti-omap (2.6.33-500.4) lucid; urgency=low [ Amit Kucheria ] * [Config] Compile-in display subsystem * Merge the DSS2 stack from 2.6.34-rc2 mainline tree * SAUCE: Upgrade aufs2 to latest version for 2.6.3

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Changed in: linux-ti-omap (Ubuntu Lucid) Status: New => Triaged -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification because you are a member of Ubuntu Bugs,

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Changed in: linux-ti-omap (Ubuntu Lucid) Importance: Undecided => High ** Changed in: linux-ti-omap (Ubuntu Lucid) Milestone: None => ubuntu-10.04-beta-2 ** Changed in: linux-ti-omap (Ubuntu Lucid) Assignee: (unassigned) => Amit Kucheria (amitk) -- VFP/NEON state is not preserved

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Also affects: linux-ti-omap (Ubuntu) Importance: Undecided Status: New -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification because you are a member of Ubu

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

This bug was fixed in the package linux-fsl-imx51 - 2.6.31-605.8 --- linux-fsl-imx51 (2.6.31-605.8) lucid; urgency=low [ Upstream Kernel Changes ] * ARM: VFP: add support to sync the VFP state of the current thread - LP: #507503 * ARM: VFP: preserve the HW context when call

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

This bug was fixed in the package linux-mvl-dove - 2.6.32-201.11 --- linux-mvl-dove (2.6.32-201.11) lucid; urgency=low [ Upstream Kernel Changes ] * ARM: VFP: add support to sync the VFP state of the current thread - LP: #507503 * ARM: VFP: preserve the HW context when call

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Changed in: linux-fsl-imx51 (Ubuntu Lucid) Milestone: lucid-alpha-3 => ubuntu-10.04-beta-1 ** Changed in: linux-mvl-dove (Ubuntu Lucid) Milestone: lucid-alpha-3 => ubuntu-10.04-beta-1 -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user pro

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

For mvl-dove kernel, patches were applied too -- linux-mvl-dove (2.6.32-201.11) lucid; urgency=low [ Upstream Kernel Changes ] * ARM: VFP: add support to sync the VFP state of the current thread - LP: #507503 * ARM: VFP: preserve the HW context when calling signal handlers - LP: #5

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

For fsl-imx51 kernel, patches were applied: -- linux-fsl-imx51 (2.6.31-605.8) lucid; urgency=low [ Upstream Kernel Changes ] * ARM: VFP: add support to sync the VFP state of the current thread - LP: #507503 * ARM: VFP: preserve the HW context when calling signal handlers - LP: #507

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

These 2 patches is in my git branch now: http://kernel.ubuntu.com/git?p=roc/ubuntu-lucid.git;a=shortlog;h=refs/heads/vfp_sig Paul and Oliver, could you please help me to test the kernel packages on your hardware: http://people.canonical.com/~roc/kernel/vfp_sig/ Thanks, -Bryan -- VFP/NEON state

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Dave, Thanks a lot. I will prepare to apply these 2 patches into our fsl- imx51 kernel Happy Chinese New Year. -Bryan ** Changed in: linux-fsl-imx51 (Ubuntu Lucid) Status: Confirmed => In Progress -- VFP/NEON state is not preserved around signal handlers, causing state corruption betw

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

FYI, someone here tries the patches referenced above (on a non-Ubuntu 2.6.33-rc5 tree I believe), and they do seem to resolve the screen corruption problem. Repeating the links here to avoid confusion: [RFC PATCH v2 0/2] ARM: VFP: Save / restore VFP state on the signal handler path

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

lowering severity some since it does not have known security implications. ** Also affects: linux-fsl-imx51 (Ubuntu Lucid) Importance: Critical Assignee: Bryan Wu (cooloney) Status: Confirmed ** Also affects: linux-mvl-dove (Ubuntu Lucid) Importance: Medium Assignee: Eric M

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

I suggest you ping the list and explain the Ubuntu timescale so that they know you're waiting for a solution. -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification becau

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Since these 2 patches are still concerned by upstream, I will wait for it's finalizing. -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification because you are a member of

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

I've had the test script running on the new kernel for several minutes now, and I never get the d0 corrupted message. Looks good! -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bu

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

I've uploaded a test kernel for dove at http://people.canonical.com/~ycmiao/vfpstate/linux- image-2.6.32-200-dove_2.6.32-200.7~vfpstate_armel.deb. -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Changed in: linux-mvl-dove (Ubuntu) Importance: Critical => Medium -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification because you are a member of Ubuntu Bugs, w

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Changed in: linux-mvl-dove (Ubuntu) Importance: Undecided => Critical ** Changed in: linux-mvl-dove (Ubuntu) Assignee: (unassigned) => Eric Miao (eric.y.miao) -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launc

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Dave, Thanks so much. I will test this on my hardware and give some feedback later. -Bryan -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification because you are a memb

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Please see the following updated patch posting to alkml by Imre Deak: [RFC PATCH v2 0/2] ARM: VFP: Save / restore VFP state on the signal handler path [RFC PATCH v2 1/2] ARM: VFP: add support to sync the VFP state of

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

The test case from the ARM Linux kernel thread does not work for me; GCC allocates the VFP use in main to d8-d15 (since main is not a leaf function). The signal handler is allocated d0-d7 to corrupt, so main does not notice. It may depend on the GCC version. To force the register allocation, I c

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Dave, Great, thanks for pushing this. So is there any test case I can run on my hardware to reproduce this issue? My X never crashed before, -;)). -Bryan -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

@Dave, I am testing these 2 patches, but how about the upstream response of that? I guess Catalin is pushing that? Thanks -Bryan -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this b

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Linking this bug to linux-mvl-dove since the problem is now known not to be imx51-specific. ** Also affects: linux-mvl-dove (Ubuntu) Importance: Undecided Status: New -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bug

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Now I understand this a bit more, it looks like there is not a security impact; just a robustness problem. The kernel does isolate the VFP/NEON state between processes, but a process can still corrupt its own state if VFP or NEON registers are modified inside a signal handler. I wrote an experime

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

Marked critical due to the potential security impact. Bryan, is the fix that was posted on the mailing list viable? Seems like there was still some debate over it. ** Changed in: linux-fsl-imx51 (Ubuntu) Importance: Undecided => Critical -- VFP/NEON state is not preserved around signal handl

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

This bug is a potential security hole and should be prioritised as such. -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.net/bugs/507503 You received this bug notification because you are a member of Ubuntu Bugs, w

[Bug 507503] Re: VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes

** Changed in: linux-fsl-imx51 (Ubuntu) Assignee: (unassigned) => Bryan Wu (cooloney) ** Changed in: linux-fsl-imx51 (Ubuntu) Status: New => Confirmed -- VFP/NEON state is not preserved around signal handlers, causing state corruption between user processes https://bugs.launchpad.ne