Hi Vinvenzo,
I'm looking through some of the remaining y2038 work, and noticed that arch/arm64 has the new generic vdso implementation, but arch/arm does.
Do you patches for using the same code on arch/arm?
Arnd
Hi Arnd,
On 10/24/19 1:34 PM, Arnd Bergmann wrote:
Hi Vinvenzo,
I'm looking through some of the remaining y2038 work, and noticed that arch/arm64 has the new generic vdso implementation, but arch/arm does.
This is correct, arch/arm does not have support for the generic vDSO library.
Do you patches for using the same code on arch/arm?
The patches were send for review together with the arm64 ones and the rest of the architectures supported but they did not get merged yet and I do not know why.
Arnd
On Thu, Oct 24, 2019 at 01:47:24PM +0100, Vincenzo Frascino wrote:
Hi Arnd,
On 10/24/19 1:34 PM, Arnd Bergmann wrote:
Hi Vinvenzo,
I'm looking through some of the remaining y2038 work, and noticed that arch/arm64 has the new generic vdso implementation, but arch/arm does.
This is correct, arch/arm does not have support for the generic vDSO library.
Do you patches for using the same code on arch/arm?
The patches were send for review together with the arm64 ones and the rest of the architectures supported but they did not get merged yet and I do not know why.
I think you've forgotten, you didn't read the email from the patch system, or the "Regression in 5.3-rc1 and later" email thread on the mailing lists.
https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=8868/1 https://lore.kernel.org/lkml/faaa3843-09a6-1a21-3448-072eeed1ea00@googlemail...
Basically, I couldn't merge for the last merge window without causing a regression, so it's delayed until the next merge window.
Thanks.
On Fri, Oct 25, 2019 at 08:17:36PM +0100, Russell King - ARM Linux admin wrote:
On Thu, Oct 24, 2019 at 01:47:24PM +0100, Vincenzo Frascino wrote:
Hi Arnd,
On 10/24/19 1:34 PM, Arnd Bergmann wrote:
Hi Vinvenzo,
I'm looking through some of the remaining y2038 work, and noticed that arch/arm64 has the new generic vdso implementation, but arch/arm does.
This is correct, arch/arm does not have support for the generic vDSO library.
Do you patches for using the same code on arch/arm?
The patches were send for review together with the arm64 ones and the rest of the architectures supported but they did not get merged yet and I do not know why.
I think you've forgotten, you didn't read the email from the patch system, or the "Regression in 5.3-rc1 and later" email thread on the mailing lists.
https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=8868/1 https://lore.kernel.org/lkml/faaa3843-09a6-1a21-3448-072eeed1ea00@googlemail...
Basically, I couldn't merge for the last merge window without causing a regression, so it's delayed until the next merge window.
Okay, having discussed with Arnd what the current situation is, he points out that he needs an additional patch to solve some further build issues with the generic VDSO patches - caused by the Kconfig changes in your first patch.
I'd rather that gets fixed up before I apply the patches, rather than applying a set of patches and a fix on top.
Please re-spin with Arnd's Kconfig changes included.
Thanks.
On Fri, Oct 25, 2019 at 9:28 PM Russell King - ARM Linux admin linux@armlinux.org.uk wrote:
On Fri, Oct 25, 2019 at 08:17:36PM +0100, Russell King - ARM Linux admin wrote:
On Thu, Oct 24, 2019 at 01:47:24PM +0100, Vincenzo Frascino wrote:
Okay, having discussed with Arnd what the current situation is, he points out that he needs an additional patch to solve some further build issues with the generic VDSO patches - caused by the Kconfig changes in your first patch.
I'd rather that gets fixed up before I apply the patches, rather than applying a set of patches and a fix on top.
Please re-spin with Arnd's Kconfig changes included.
For reference, this is the change I needed to get randconfig builds working in configurations that did not enable CONFIG_VDSO:
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index 32184065ceda..776ac0f4ef0e 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -60,8 +60,6 @@ config ARM select GENERIC_SMP_IDLE_THREAD select GENERIC_STRNCPY_FROM_USER select GENERIC_STRNLEN_USER - select GENERIC_GETTIMEOFDAY - select GENERIC_VDSO_32 select HANDLE_DOMAIN_IRQ select HARDIRQS_SW_RESEND select HAVE_ARCH_AUDITSYSCALL if AEABI && !OABI_COMPAT @@ -112,7 +110,6 @@ config ARM select HAVE_SYSCALL_TRACEPOINTS select HAVE_UID16 select HAVE_VIRT_CPU_ACCOUNTING_GEN - select HAVE_GENERIC_VDSO if AEABI select IRQ_FORCED_THREADING select MODULES_USE_ELF_REL select NEED_DMA_MAP_STATE diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig index 608a1f6414c2..dce67c6834a8 100644 --- a/arch/arm/mm/Kconfig +++ b/arch/arm/mm/Kconfig @@ -896,7 +896,10 @@ config VDSO bool "Enable VDSO for acceleration of some system calls" depends on AEABI && MMU && CPU_V7 default y if ARM_ARCH_TIMER + select HAVE_GENERIC_VDSO select GENERIC_TIME_VSYSCALL + select GENERIC_VDSO_32 + select GENERIC_GETTIMEOFDAY help Place in the process address space an ELF shared object providing fast implementations of gettimeofday and
Hi Russell,
On 10/25/19 8:28 PM, Russell King - ARM Linux admin wrote:
On Fri, Oct 25, 2019 at 08:17:36PM +0100, Russell King - ARM Linux admin wrote:
On Thu, Oct 24, 2019 at 01:47:24PM +0100, Vincenzo Frascino wrote:
Hi Arnd,
On 10/24/19 1:34 PM, Arnd Bergmann wrote:
Hi Vinvenzo,
I'm looking through some of the remaining y2038 work, and noticed that arch/arm64 has the new generic vdso implementation, but arch/arm does.
This is correct, arch/arm does not have support for the generic vDSO library.
Do you patches for using the same code on arch/arm?
The patches were send for review together with the arm64 ones and the rest of the architectures supported but they did not get merged yet and I do not know why.
I think you've forgotten, you didn't read the email from the patch system, or the "Regression in 5.3-rc1 and later" email thread on the mailing lists.
https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=8868/1 https://lore.kernel.org/lkml/faaa3843-09a6-1a21-3448-072eeed1ea00@googlemail...
Basically, I couldn't merge for the last merge window without causing a regression, so it's delayed until the next merge window.
Okay, having discussed with Arnd what the current situation is, he points out that he needs an additional patch to solve some further build issues with the generic VDSO patches - caused by the Kconfig changes in your first patch.
I'd rather that gets fixed up before I apply the patches, rather than applying a set of patches and a fix on top.
Please re-spin with Arnd's Kconfig changes included.
No problem, I will re-spin the patches.
Thanks.
Hi Arnd,
On 10/25/19 8:33 PM, Arnd Bergmann wrote:
On Fri, Oct 25, 2019 at 9:28 PM Russell King - ARM Linux admin linux@armlinux.org.uk wrote:
On Fri, Oct 25, 2019 at 08:17:36PM +0100, Russell King - ARM Linux admin wrote:
On Thu, Oct 24, 2019 at 01:47:24PM +0100, Vincenzo Frascino wrote:
Okay, having discussed with Arnd what the current situation is, he points out that he needs an additional patch to solve some further build issues with the generic VDSO patches - caused by the Kconfig changes in your first patch.
I'd rather that gets fixed up before I apply the patches, rather than applying a set of patches and a fix on top.
Please re-spin with Arnd's Kconfig changes included.
For reference, this is the change I needed to get randconfig builds working in configurations that did not enable CONFIG_VDSO:
Thanks for this.
-
Arnd Bergmann -
Russell King - ARM Linux admin -
Vincenzo Frascino