From: Guo Jin <[email protected]>
When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y
and CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0,
an error: "<inline asm> error: changed section flags for
.spinlock.text, expected:: 0x6" occurred.
The reason is the .spinlock.text in kernel/locking/qspinlock.o
is used many times, but its flags are omitted in subsequent use.
LLVM 11.0 assembler didn't permit to
leave out flags in subsequent uses of the same sections.
So this patch adds the corresponding flags to avoid above error.
Signed-off-by: Guo Jin <[email protected]>
---
arch/x86/include/asm/qspinlock_paravirt.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/include/asm/qspinlock_paravirt.h b/arch/x86/include/asm/qspinlock_paravirt.h
index 60ece592b220..dbb38a6b4dfb 100644
--- a/arch/x86/include/asm/qspinlock_paravirt.h
+++ b/arch/x86/include/asm/qspinlock_paravirt.h
@@ -37,7 +37,7 @@ __PV_CALLEE_SAVE_REGS_THUNK(__pv_queued_spin_unlock_slowpath, ".spinlock.text");
* rsi = lockval (second argument)
* rdx = internal variable (set to 0)
*/
-asm (".pushsection .spinlock.text;"
+asm (".pushsection .spinlock.text, \"ax\";"
".globl " PV_UNLOCK ";"
".type " PV_UNLOCK ", @function;"
".align 4,0x90;"
--
2.25.1
On Tue, Nov 08, 2022 at 02:01:26PM +0800, [email protected] wrote:
> From: Guo Jin <[email protected]>
>
> When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y
> and CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0,
> an error: "<inline asm> error: changed section flags for
> .spinlock.text, expected:: 0x6" occurred.
>
> The reason is the .spinlock.text in kernel/locking/qspinlock.o
> is used many times, but its flags are omitted in subsequent use.
>
> LLVM 11.0 assembler didn't permit to
> leave out flags in subsequent uses of the same sections.
>
> So this patch adds the corresponding flags to avoid above error.
>
> Signed-off-by: Guo Jin <[email protected]>
Right, my reverse bisect tells me that this error was fixed by [1] in
LLVM, which landed in 12.0.0 and was backported to LLVM 11.0.1 [2].
We could bump the minimum supported version of LLVM for building the
kernel to 11.0.1 over this, just to avoid further issues like this down
the road, as the kernel will always be adding new sections for various
things. That change would not appear to be disruptive, as all the major
distros either have 11.0.1 or newer or older than the current minimum of
11.0.0, so they need to get clang from elsewhere already.
archlinux:latest: clang version 14.0.6
debian:oldoldstable-slim: clang version 3.8.1-24 (tags/RELEASE_381/final)
debian:oldstable-slim: clang version 7.0.1-8+deb10u2 (tags/RELEASE_701/final)
debian:stable-slim: Debian clang version 11.0.1-2
debian:testing-slim: Debian clang version 14.0.6-2
debian:unstable-slim: Debian clang version 14.0.6-7
fedora:35: clang version 13.0.1 (Fedora 13.0.1-1.fc35)
fedora:latest: clang version 14.0.5 (Fedora 14.0.5-1.fc36)
fedora:rawhide: clang version 15.0.4 (Fedora 15.0.4-1.fc38)
opensuse/leap:15: clang version 13.0.1
opensuse/leap:latest: clang version 13.0.1
opensuse/tumbleweed:latest: clang version 15.0.3
ubuntu:bionic: clang version 6.0.0-1ubuntu2 (tags/RELEASE_600/final)
ubuntu:focal: clang version 10.0.0-4ubuntu1
ubuntu:latest: Ubuntu clang version 14.0.0-1ubuntu1
ubuntu:kinetic: Ubuntu clang version 15.0.2-1
ubuntu:rolling: Ubuntu clang version 15.0.2-1
ubuntu:devel: Ubuntu clang version 15.0.2-1
At the same time, you are clearly using LLVM 11.0.0, so I think this is a
reasonable fix in the immediate term.
Reviewed-by: Nathan Chancellor <[email protected]>
[1]: https://github.com/llvm/llvm-project/commit/1deff4009e0ae661b03682901bf6932297ce7ea1
[2]: https://github.com/llvm/llvm-project/commit/700baa009dc685a0adc5f94d258be4ae24742471
> ---
> arch/x86/include/asm/qspinlock_paravirt.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/include/asm/qspinlock_paravirt.h b/arch/x86/include/asm/qspinlock_paravirt.h
> index 60ece592b220..dbb38a6b4dfb 100644
> --- a/arch/x86/include/asm/qspinlock_paravirt.h
> +++ b/arch/x86/include/asm/qspinlock_paravirt.h
> @@ -37,7 +37,7 @@ __PV_CALLEE_SAVE_REGS_THUNK(__pv_queued_spin_unlock_slowpath, ".spinlock.text");
> * rsi = lockval (second argument)
> * rdx = internal variable (set to 0)
> */
> -asm (".pushsection .spinlock.text;"
> +asm (".pushsection .spinlock.text, \"ax\";"
> ".globl " PV_UNLOCK ";"
> ".type " PV_UNLOCK ", @function;"
> ".align 4,0x90;"
> --
> 2.25.1
>
Thanks your comments,
And if any concern to apply it on upstream?
Yadi
-----?ʼ?ԭ??-----
??????: [email protected] <[email protected]>
????ʱ??: 2022??11??9?? 6:40
?ռ???: [email protected]
????: [email protected]; [email protected]; [email protected];
[email protected]; [email protected]; [email protected];
[email protected]; [email protected]; [email protected]; linux-kernel@vger.
kernel.org; [email protected]; [email protected]; [email protected];
[email protected]; [email protected]
????: Re: [PATCH] locking: fix kernel/locking/ inline asm error
On Tue, Nov 08, 2022 at 02:01:26PM +0800, [email protected] wrote:
> From: Guo Jin <[email protected]>
>
> When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y and
> CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0, an error:
> "<inline asm> error: changed section flags for .spinlock.text,
> expected:: 0x6" occurred.
>
> The reason is the .spinlock.text in kernel/locking/qspinlock.o is used
> many times, but its flags are omitted in subsequent use.
>
> LLVM 11.0 assembler didn't permit to
> leave out flags in subsequent uses of the same sections.
>
> So this patch adds the corresponding flags to avoid above error.
>
> Signed-off-by: Guo Jin <[email protected]>
Right, my reverse bisect tells me that this error was fixed by [1] in LLVM,
which landed in 12.0.0 and was backported to LLVM 11.0.1 [2].
We could bump the minimum supported version of LLVM for building the kernel
to 11.0.1 over this, just to avoid further issues like this down the road,
as the kernel will always be adding new sections for various things. That
change would not appear to be disruptive, as all the major distros either
have 11.0.1 or newer or older than the current minimum of 11.0.0, so they
need to get clang from elsewhere already.
archlinux:latest: clang version 14.0.6
debian:oldoldstable-slim: clang version 3.8.1-24 (tags/RELEASE_381/final)
debian:oldstable-slim: clang version 7.0.1-8+deb10u2
(tags/RELEASE_701/final)
debian:stable-slim: Debian clang version 11.0.1-2
debian:testing-slim: Debian clang version 14.0.6-2
debian:unstable-slim: Debian clang version 14.0.6-7
fedora:35: clang version 13.0.1 (Fedora 13.0.1-1.fc35)
fedora:latest: clang version 14.0.5 (Fedora 14.0.5-1.fc36)
fedora:rawhide: clang version 15.0.4 (Fedora 15.0.4-1.fc38)
opensuse/leap:15: clang version 13.0.1
opensuse/leap:latest: clang version 13.0.1
opensuse/tumbleweed:latest: clang version 15.0.3
ubuntu:bionic: clang version 6.0.0-1ubuntu2 (tags/RELEASE_600/final)
ubuntu:focal: clang version 10.0.0-4ubuntu1
ubuntu:latest: Ubuntu clang version 14.0.0-1ubuntu1
ubuntu:kinetic: Ubuntu clang version 15.0.2-1
ubuntu:rolling: Ubuntu clang version 15.0.2-1
ubuntu:devel: Ubuntu clang version 15.0.2-1
At the same time, you are clearly using LLVM 11.0.0, so I think this is a
reasonable fix in the immediate term.
Reviewed-by: Nathan Chancellor <[email protected]>
[1]:
https://github.com/llvm/llvm-project/commit/1deff4009e0ae661b03682901bf69322
97ce7ea1
[2]:
https://github.com/llvm/llvm-project/commit/700baa009dc685a0adc5f94d258be4ae
24742471
> ---
> arch/x86/include/asm/qspinlock_paravirt.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/include/asm/qspinlock_paravirt.h
> b/arch/x86/include/asm/qspinlock_paravirt.h
> index 60ece592b220..dbb38a6b4dfb 100644
> --- a/arch/x86/include/asm/qspinlock_paravirt.h
> +++ b/arch/x86/include/asm/qspinlock_paravirt.h
> @@ -37,7 +37,7 @@
__PV_CALLEE_SAVE_REGS_THUNK(__pv_queued_spin_unlock_slowpath,
".spinlock.text");
> * rsi = lockval (second argument)
> * rdx = internal variable (set to 0)
> */
> -asm (".pushsection .spinlock.text;"
> +asm (".pushsection .spinlock.text, \"ax\";"
> ".globl " PV_UNLOCK ";"
> ".type " PV_UNLOCK ", @function;"
> ".align 4,0x90;"
> --
> 2.25.1
>
On Thu, Nov 10, 2022 at 06:05:13PM +0800, [email protected] wrote:
>
> Thanks your comments,
> And if any concern to apply it on upstream?
Unfortunately, I am not the one who gets to make that decision; one of
the maintainers of this file will have to pick it up, assuming they see
no issue with it.
Cheers,
Nathan
> -----邮件原件-----
> 发件人: [email protected] <[email protected]>
> 发送时间: 2022年11月9日 6:40
> 收件人: [email protected]
> 抄送: [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; [email protected]; linux-kernel@vger.
> kernel.org; [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]
> 主题: Re: [PATCH] locking: fix kernel/locking/ inline asm error
>
> On Tue, Nov 08, 2022 at 02:01:26PM +0800, [email protected] wrote:
> > From: Guo Jin <[email protected]>
> >
> > When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y and
> > CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0, an error:
> > "<inline asm> error: changed section flags for .spinlock.text,
> > expected:: 0x6" occurred.
> >
> > The reason is the .spinlock.text in kernel/locking/qspinlock.o is used
> > many times, but its flags are omitted in subsequent use.
> >
> > LLVM 11.0 assembler didn't permit to
> > leave out flags in subsequent uses of the same sections.
> >
> > So this patch adds the corresponding flags to avoid above error.
> >
> > Signed-off-by: Guo Jin <[email protected]>
>
> Right, my reverse bisect tells me that this error was fixed by [1] in LLVM,
> which landed in 12.0.0 and was backported to LLVM 11.0.1 [2].
>
> We could bump the minimum supported version of LLVM for building the kernel
> to 11.0.1 over this, just to avoid further issues like this down the road,
> as the kernel will always be adding new sections for various things. That
> change would not appear to be disruptive, as all the major distros either
> have 11.0.1 or newer or older than the current minimum of 11.0.0, so they
> need to get clang from elsewhere already.
>
> archlinux:latest: clang version 14.0.6
>
> debian:oldoldstable-slim: clang version 3.8.1-24 (tags/RELEASE_381/final)
> debian:oldstable-slim: clang version 7.0.1-8+deb10u2
> (tags/RELEASE_701/final)
> debian:stable-slim: Debian clang version 11.0.1-2
> debian:testing-slim: Debian clang version 14.0.6-2
> debian:unstable-slim: Debian clang version 14.0.6-7
>
> fedora:35: clang version 13.0.1 (Fedora 13.0.1-1.fc35)
> fedora:latest: clang version 14.0.5 (Fedora 14.0.5-1.fc36)
> fedora:rawhide: clang version 15.0.4 (Fedora 15.0.4-1.fc38)
>
> opensuse/leap:15: clang version 13.0.1
> opensuse/leap:latest: clang version 13.0.1
> opensuse/tumbleweed:latest: clang version 15.0.3
>
> ubuntu:bionic: clang version 6.0.0-1ubuntu2 (tags/RELEASE_600/final)
> ubuntu:focal: clang version 10.0.0-4ubuntu1
> ubuntu:latest: Ubuntu clang version 14.0.0-1ubuntu1
> ubuntu:kinetic: Ubuntu clang version 15.0.2-1
> ubuntu:rolling: Ubuntu clang version 15.0.2-1
> ubuntu:devel: Ubuntu clang version 15.0.2-1
>
> At the same time, you are clearly using LLVM 11.0.0, so I think this is a
> reasonable fix in the immediate term.
>
> Reviewed-by: Nathan Chancellor <[email protected]>
>
> [1]:
> https://github.com/llvm/llvm-project/commit/1deff4009e0ae661b03682901bf69322
> 97ce7ea1
> [2]:
> https://github.com/llvm/llvm-project/commit/700baa009dc685a0adc5f94d258be4ae
> 24742471
>
> > ---
> > arch/x86/include/asm/qspinlock_paravirt.h | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/arch/x86/include/asm/qspinlock_paravirt.h
> > b/arch/x86/include/asm/qspinlock_paravirt.h
> > index 60ece592b220..dbb38a6b4dfb 100644
> > --- a/arch/x86/include/asm/qspinlock_paravirt.h
> > +++ b/arch/x86/include/asm/qspinlock_paravirt.h
> > @@ -37,7 +37,7 @@
> __PV_CALLEE_SAVE_REGS_THUNK(__pv_queued_spin_unlock_slowpath,
> ".spinlock.text");
> > * rsi = lockval (second argument)
> > * rdx = internal variable (set to 0)
> > */
> > -asm (".pushsection .spinlock.text;"
> > +asm (".pushsection .spinlock.text, \"ax\";"
> > ".globl " PV_UNLOCK ";"
> > ".type " PV_UNLOCK ", @function;"
> > ".align 4,0x90;"
> > --
> > 2.25.1
> >
>
The following commit has been merged into the locking/urgent branch of tip:
Commit-ID: 23df39fc6a36183af5e6e4f47523f1ad2cdc1d30
Gitweb: https://git.kernel.org/tip/23df39fc6a36183af5e6e4f47523f1ad2cdc1d30
Author: Guo Jin <[email protected]>
AuthorDate: Tue, 08 Nov 2022 14:01:26 +08:00
Committer: Peter Zijlstra <[email protected]>
CommitterDate: Wed, 16 Nov 2022 10:18:09 +01:00
locking: Fix qspinlock/x86 inline asm error
When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y and
CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0, an error:
"<inline asm> error: changed section flags for .spinlock.text,
expected:: 0x6" occurred.
The reason is the .spinlock.text in kernel/locking/qspinlock.o
is used many times, but its flags are omitted in subsequent use.
LLVM 11.0 assembler didn't permit to
leave out flags in subsequent uses of the same sections.
So this patch adds the corresponding flags to avoid above error.
Fixes: 501f7f69bca1 ("locking: Add __lockfunc to slow path functions")
Signed-off-by: Guo Jin <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Nathan Chancellor <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
arch/x86/include/asm/qspinlock_paravirt.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/include/asm/qspinlock_paravirt.h b/arch/x86/include/asm/qspinlock_paravirt.h
index 60ece59..dbb38a6 100644
--- a/arch/x86/include/asm/qspinlock_paravirt.h
+++ b/arch/x86/include/asm/qspinlock_paravirt.h
@@ -37,7 +37,7 @@ __PV_CALLEE_SAVE_REGS_THUNK(__pv_queued_spin_unlock_slowpath, ".spinlock.text");
* rsi = lockval (second argument)
* rdx = internal variable (set to 0)
*/
-asm (".pushsection .spinlock.text;"
+asm (".pushsection .spinlock.text, \"ax\";"
".globl " PV_UNLOCK ";"
".type " PV_UNLOCK ", @function;"
".align 4,0x90;"
On 16.11.22 10:21, tip-bot2 for Guo Jin wrote:
> The following commit has been merged into the locking/urgent branch of tip:
>
> Commit-ID: 23df39fc6a36183af5e6e4f47523f1ad2cdc1d30
> Gitweb: https://git.kernel.org/tip/23df39fc6a36183af5e6e4f47523f1ad2cdc1d30
> Author: Guo Jin <[email protected]>
> AuthorDate: Tue, 08 Nov 2022 14:01:26 +08:00
> Committer: Peter Zijlstra <[email protected]>
> CommitterDate: Wed, 16 Nov 2022 10:18:09 +01:00
>
> locking: Fix qspinlock/x86 inline asm error
>
> When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y and
> CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0, an error:
> "<inline asm> error: changed section flags for .spinlock.text,
> expected:: 0x6" occurred.
>
> The reason is the .spinlock.text in kernel/locking/qspinlock.o
> is used many times, but its flags are omitted in subsequent use.
>
> LLVM 11.0 assembler didn't permit to
> leave out flags in subsequent uses of the same sections.
>
> So this patch adds the corresponding flags to avoid above error.
>
> Fixes: 501f7f69bca1 ("locking: Add __lockfunc to slow path functions")
> Signed-off-by: Guo Jin <[email protected]>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
> Reviewed-by: Nathan Chancellor <[email protected]>
> Link: https://lore.kernel.org/r/[email protected]
Wouldn't it be better to take my more generic patch [1] instead?
[1]: https://lore.kernel.org/lkml/[email protected]/
Juergen
On Wed, Nov 16, 2022 at 10:40:10AM +0100, Juergen Gross wrote:
> On 16.11.22 10:21, tip-bot2 for Guo Jin wrote:
> > The following commit has been merged into the locking/urgent branch of tip:
> >
> > Commit-ID: 23df39fc6a36183af5e6e4f47523f1ad2cdc1d30
> > Gitweb: https://git.kernel.org/tip/23df39fc6a36183af5e6e4f47523f1ad2cdc1d30
> > Author: Guo Jin <[email protected]>
> > AuthorDate: Tue, 08 Nov 2022 14:01:26 +08:00
> > Committer: Peter Zijlstra <[email protected]>
> > CommitterDate: Wed, 16 Nov 2022 10:18:09 +01:00
> >
> > locking: Fix qspinlock/x86 inline asm error
> >
> > When compiling linux 6.1.0-rc3 configured with CONFIG_64BIT=y and
> > CONFIG_PARAVIRT_SPINLOCKS=y on x86_64 using LLVM 11.0, an error:
> > "<inline asm> error: changed section flags for .spinlock.text,
> > expected:: 0x6" occurred.
> >
> > The reason is the .spinlock.text in kernel/locking/qspinlock.o
> > is used many times, but its flags are omitted in subsequent use.
> >
> > LLVM 11.0 assembler didn't permit to
> > leave out flags in subsequent uses of the same sections.
> >
> > So this patch adds the corresponding flags to avoid above error.
> >
> > Fixes: 501f7f69bca1 ("locking: Add __lockfunc to slow path functions")
> > Signed-off-by: Guo Jin <[email protected]>
> > Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
> > Reviewed-by: Nathan Chancellor <[email protected]>
> > Link: https://lore.kernel.org/r/[email protected]
>
> Wouldn't it be better to take my more generic patch [1] instead?
>
> [1]: https://lore.kernel.org/lkml/[email protected]/
Durr, I missed that, lemme go stare at it.