| Message ID | 20200715125614.3240269-1-maz@kernel.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | arm64: Allow the compat vdso to be disabled at runtime | expand |
On Wed, Jul 15, 2020 at 01:56:11PM +0100, Marc Zyngier wrote: > This is a backport of the series that recently went into 5.8. Note > that the first patch is more a complete rewriting than a backport, as > the vdso implementation in 5.4 doesn't have much in common with > mainline. This affects the 32bit arch code in a benign way. > > It has seen very little testing, as I don't have the HW that triggers > this issue. I have run it in VMs by faking the CPU MIDR, and nothing > caught fire. Famous last words. These are also needed in 5.7.y, right? If so, I need that series before I can take this one as we don't want people moving to a newer kernel and suffer regressions :( thanks, greg k-h
Hi Greg, On 2020-07-16 12:58, Greg KH wrote: > On Wed, Jul 15, 2020 at 01:56:11PM +0100, Marc Zyngier wrote: >> This is a backport of the series that recently went into 5.8. Note >> that the first patch is more a complete rewriting than a backport, as >> the vdso implementation in 5.4 doesn't have much in common with >> mainline. This affects the 32bit arch code in a benign way. >> >> It has seen very little testing, as I don't have the HW that triggers >> this issue. I have run it in VMs by faking the CPU MIDR, and nothing >> caught fire. Famous last words. > > These are also needed in 5.7.y, right? If so, I need that series > before > I can take this one as we don't want people moving to a newer kernel > and > suffer regressions :( The original mainline changes: 4b661d6133c5 arm64: arch_timer: Disable the compat vdso for cores affected by ARM64_WORKAROUND_1418040 c1fbec4ac0d7 arm64: arch_timer: Allow an workaround descriptor to disable compat vdso 97884ca8c292 arm64: Introduce a way to disable the 32bit vdso do apply cleanly to stable-5.7. Do you want me to resend them separately, or will you pick the patches directly from mainline? Thanks, M.
On Fri, Jul 17, 2020 at 09:02:06AM +0100, Marc Zyngier wrote: > Hi Greg, > > On 2020-07-16 12:58, Greg KH wrote: > > On Wed, Jul 15, 2020 at 01:56:11PM +0100, Marc Zyngier wrote: > > > This is a backport of the series that recently went into 5.8. Note > > > that the first patch is more a complete rewriting than a backport, as > > > the vdso implementation in 5.4 doesn't have much in common with > > > mainline. This affects the 32bit arch code in a benign way. > > > > > > It has seen very little testing, as I don't have the HW that triggers > > > this issue. I have run it in VMs by faking the CPU MIDR, and nothing > > > caught fire. Famous last words. > > > > These are also needed in 5.7.y, right? If so, I need that series before > > I can take this one as we don't want people moving to a newer kernel and > > suffer regressions :( > > The original mainline changes: > > 4b661d6133c5 arm64: arch_timer: Disable the compat vdso for cores affected > by ARM64_WORKAROUND_1418040 > c1fbec4ac0d7 arm64: arch_timer: Allow an workaround descriptor to disable > compat vdso > 97884ca8c292 arm64: Introduce a way to disable the 32bit vdso > > do apply cleanly to stable-5.7. Do you want me to resend them separately, > or will you pick the patches directly from mainline? Hm, cherry-pick seems to work due to file renames, let me try this again...
On Fri, Jul 17, 2020 at 11:33:57AM +0200, Greg KH wrote: > On Fri, Jul 17, 2020 at 09:02:06AM +0100, Marc Zyngier wrote: > > Hi Greg, > > > > On 2020-07-16 12:58, Greg KH wrote: > > > On Wed, Jul 15, 2020 at 01:56:11PM +0100, Marc Zyngier wrote: > > > > This is a backport of the series that recently went into 5.8. Note > > > > that the first patch is more a complete rewriting than a backport, as > > > > the vdso implementation in 5.4 doesn't have much in common with > > > > mainline. This affects the 32bit arch code in a benign way. > > > > > > > > It has seen very little testing, as I don't have the HW that triggers > > > > this issue. I have run it in VMs by faking the CPU MIDR, and nothing > > > > caught fire. Famous last words. > > > > > > These are also needed in 5.7.y, right? If so, I need that series before > > > I can take this one as we don't want people moving to a newer kernel and > > > suffer regressions :( > > > > The original mainline changes: > > > > 4b661d6133c5 arm64: arch_timer: Disable the compat vdso for cores affected > > by ARM64_WORKAROUND_1418040 > > c1fbec4ac0d7 arm64: arch_timer: Allow an workaround descriptor to disable > > compat vdso > > 97884ca8c292 arm64: Introduce a way to disable the 32bit vdso > > > > do apply cleanly to stable-5.7. Do you want me to resend them separately, > > or will you pick the patches directly from mainline? > > Hm, cherry-pick seems to work due to file renames, let me try this > again... Ok, my fault, these are already all in 5.7.9, sorry for the noise. I'll go queue these up now. greg "I need more coffee..." k-h
This is a backport of the series that recently went into 5.8. Note that the first patch is more a complete rewriting than a backport, as the vdso implementation in 5.4 doesn't have much in common with mainline. This affects the 32bit arch code in a benign way. It has seen very little testing, as I don't have the HW that triggers this issue. I have run it in VMs by faking the CPU MIDR, and nothing caught fire. Famous last words. The original cover letter follows. M. The relatively recent introduction of the compat vdso on arm64 has overlooked its interactions with some of the interesting errata workarounds, such as ARM64_ERRATUM_1418040 (and its older 1188873 incarnation). This erratum requires the 64bit kernel to trap 32bit accesses to the virtual counter and emulate it. When the workaround was introduced, the compat vdso simply wasn't a thing. Now that the patches have landed in mainline, we trap the CVTVCT accesses from the vdso. This can end-up in a nasty loop in the vdso, where the sequence number changes on each trap, never stabilising, and leaving userspace in a bit of a funny state (which is why we disable the vdso in most similar cases). This erratum mentionned above is a bit special in the sense that in only requires to trap AArch32 accesses, and 64bit tasks can be left alone. Consequently, the vdso is never disabled and AArch32 tasks are affected. Obviously, we really want to retain the 64bit vdso in this case. To that effect, this series offers a way to disable the 32bit view of the vdso without impacting its 64bit counterpart, by providing a "no-compat" vdso clock_mode, and plugging this feature into the 1418040 detection code. Marc Zyngier (3): arm64: Introduce a way to disable the 32bit vdso arm64: arch_timer: Allow an workaround descriptor to disable compat vdso arm64: arch_timer: Disable the compat vdso for cores affected by ARM64_WORKAROUND_1418040 arch/arm/include/asm/clocksource.h | 11 ++++++++++- arch/arm/kernel/vdso.c | 2 +- arch/arm64/include/asm/arch_timer.h | 1 + arch/arm64/include/asm/clocksource.h | 5 ++++- arch/arm64/include/asm/vdso/clocksource.h | 14 ++++++++++++++ .../include/asm/vdso/compat_gettimeofday.h | 5 +++-- arch/arm64/include/asm/vdso/gettimeofday.h | 6 ++++-- arch/arm64/include/asm/vdso/vsyscall.h | 4 +--- drivers/clocksource/arm_arch_timer.c | 19 +++++++++++++++---- 9 files changed, 53 insertions(+), 14 deletions(-) create mode 100644 arch/arm64/include/asm/vdso/clocksource.h