* backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
@ 2022-10-24 21:06 Nick Desaulniers
2022-10-25 16:53 ` Nathan Chancellor
2022-10-27 10:33 ` Greg KH
0 siblings, 2 replies; 4+ messages in thread
From: Nick Desaulniers @ 2022-10-24 21:06 UTC (permalink / raw)
To: Greg KH, Sasha Levin
Cc: Masahiro Yamada, # 3.4.x, Nathan Chancellor, clang-built-linux,
Alexey Alexandrov, Bill Wendling, Greg Thelen
[-- Attachment #1: Type: text/plain, Size: 536 bytes --]
Dear stable kernel maintainers,
Our production kernel team and ChromeOS kernel teams are reporting
that they are unable to symbolize addresses of symbols defined in
assembly sources due to a regression I caused with
commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
I fixed this upstream with
commit 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
but I think this is infeasible to backport through to 4.19.y.
Do the attached branch-specific variants look acceptable?
--
Thanks,
~Nick Desaulniers
[-- Attachment #2: 61f2b7c7497b.5.4.patch --]
[-- Type: application/octet-stream, Size: 1468 bytes --]
From b9b910ec63355521eb81f92d01b7362640d27dce Mon Sep 17 00:00:00 2001
From: Nick Desaulniers <ndesaulniers@google.com>
Date: Mon, 24 Oct 2022 13:34:14 -0700
Subject: [PATCH 5.4.y stable only] Makefile.debug: re-enable debug info for .S
files
This is _not_ an upstream commit and just for 5.4.y only. It is based
on
commit 32ef9e5054ec0321b9336058c58ec749e9c6b0fe upstream.
Alexey reported that the fraction of unknown filename instances in
kallsyms grew from ~0.3% to ~10% recently; Bill and Greg tracked it down
to assembler defined symbols, which regressed as a result of:
commit b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
In that commit, I allude to restoring debug info for assembler defined
symbols in a follow up patch, but it seems I forgot to do so in
commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
Fixes: b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
---
Makefile | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/Makefile b/Makefile
index 58325c33ef0c..ee7f818bcd99 100644
--- a/Makefile
+++ b/Makefile
@@ -802,7 +802,9 @@ DEBUG_CFLAGS += -gsplit-dwarf
else
DEBUG_CFLAGS += -g
endif
-ifneq ($(LLVM_IAS),1)
+ifeq ($(LLVM_IAS),1)
+KBUILD_AFLAGS += -g
+else
KBUILD_AFLAGS += -Wa,-gdwarf-2
endif
endif
base-commit: fd92cfed8bc6668d314acd1e6da708a80826f768
--
2.38.0.135.g90850a2211-goog
[-- Attachment #3: 61f2b7c7497b.5.10.patch --]
[-- Type: application/octet-stream, Size: 1436 bytes --]
From ae132888c0bb3c16e5b6a67709a781ba40478ac3 Mon Sep 17 00:00:00 2001
From: Nick Desaulniers <ndesaulniers@google.com>
Date: Mon, 24 Oct 2022 13:34:14 -0700
Subject: [PATCH 5.10.y stable only] Makefile.debug: re-enable debug info for
.S files
This is _not_ an upstream commit and just for 5.10.y only. It is based
on
commit 32ef9e5054ec0321b9336058c58ec749e9c6b0fe upstream.
Alexey reported that the fraction of unknown filename instances in
kallsyms grew from ~0.3% to ~10% recently; Bill and Greg tracked it down
to assembler defined symbols, which regressed as a result of:
commit b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
In that commit, I allude to restoring debug info for assembler defined
symbols in a follow up patch, but it seems I forgot to do so in
commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
Fixes: b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
---
Makefile | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/Makefile b/Makefile
index b824bdb0457c..82a7b661c1fe 100644
--- a/Makefile
+++ b/Makefile
@@ -839,7 +839,9 @@ else
DEBUG_CFLAGS += -g
endif
-ifneq ($(LLVM_IAS),1)
+ifeq ($(LLVM_IAS),1)
+KBUILD_AFLAGS += -g
+else
KBUILD_AFLAGS += -Wa,-gdwarf-2
endif
base-commit: 09be132bfe3a3075ddf160cc75865370ea35a0aa
--
2.38.0.135.g90850a2211-goog
[-- Attachment #4: 61f2b7c7497b.5.15.patch --]
[-- Type: application/octet-stream, Size: 1442 bytes --]
From 8b4e6e028ecfab0639b6fe38d8a108f510744308 Mon Sep 17 00:00:00 2001
From: Nick Desaulniers <ndesaulniers@google.com>
Date: Mon, 24 Oct 2022 13:34:14 -0700
Subject: [PATCH 5.15.y stable only] Makefile.debug: re-enable debug info for
.S files
This is _not_ an upstream commit and just for 5.15.y only. It is based
on
commit 32ef9e5054ec0321b9336058c58ec749e9c6b0fe upstream.
Alexey reported that the fraction of unknown filename instances in
kallsyms grew from ~0.3% to ~10% recently; Bill and Greg tracked it down
to assembler defined symbols, which regressed as a result of:
commit b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
In that commit, I allude to restoring debug info for assembler defined
symbols in a follow up patch, but it seems I forgot to do so in
commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
Fixes: b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
---
Makefile | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/Makefile b/Makefile
index 86b6ca862e39..9f6eb5581576 100644
--- a/Makefile
+++ b/Makefile
@@ -870,7 +870,9 @@ else
DEBUG_CFLAGS += -g
endif
-ifndef CONFIG_AS_IS_LLVM
+ifdef CONFIG_AS_IS_LLVM
+KBUILD_AFLAGS += -g
+else
KBUILD_AFLAGS += -Wa,-gdwarf-2
endif
base-commit: a3f2f5ac9d61e973e383f17a95cf2aa384e2d0c4
--
2.38.0.135.g90850a2211-goog
[-- Attachment #5: 61f2b7c7497b.4.19.patch --]
[-- Type: application/octet-stream, Size: 1497 bytes --]
From 0d1685c2e8de1b8a3a8ea3184b024544c3e2dc09 Mon Sep 17 00:00:00 2001
From: Nick Desaulniers <ndesaulniers@google.com>
Date: Mon, 24 Oct 2022 13:34:14 -0700
Subject: [PATCH 4.19.y stable only] Makefile.debug: re-enable debug info for
.S files
This is _not_ an upstream commit and just for 4.19.y only. It is based
on
commit 32ef9e5054ec0321b9336058c58ec749e9c6b0fe upstream.
Alexey reported that the fraction of unknown filename instances in
kallsyms grew from ~0.3% to ~10% recently; Bill and Greg tracked it down
to assembler defined symbols, which regressed as a result of:
commit b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
In that commit, I allude to restoring debug info for assembler defined
symbols in a follow up patch, but it seems I forgot to do so in
commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
Fixes: b8a9092330da ("Kbuild: do not emit debug info for assembly with LLVM_IAS=1")
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
---
Makefile | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/Makefile b/Makefile
index 165812d2f786..1471619f50ca 100644
--- a/Makefile
+++ b/Makefile
@@ -744,7 +744,9 @@ KBUILD_CFLAGS += $(call cc-option, -gsplit-dwarf, -g)
else
KBUILD_CFLAGS += -g
endif
-ifneq ($(LLVM_IAS),1)
+ifeq ($(LLVM_IAS),1)
+KBUILD_AFLAGS += -g
+else
KBUILD_AFLAGS += -Wa,-gdwarf-2
endif
endif
base-commit: cf46ee80c6d2f24c2cae2b40c7d45b6e81457b8b
--
2.38.0.135.g90850a2211-goog
^ permalink raw reply related [flat|nested] 4+ messages in thread* Re: backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
2022-10-24 21:06 backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files") Nick Desaulniers
@ 2022-10-25 16:53 ` Nathan Chancellor
2022-10-26 17:50 ` Nick Desaulniers
2022-10-27 10:33 ` Greg KH
1 sibling, 1 reply; 4+ messages in thread
From: Nathan Chancellor @ 2022-10-25 16:53 UTC (permalink / raw)
To: Nick Desaulniers
Cc: Greg KH, Sasha Levin, Masahiro Yamada, # 3.4.x,
clang-built-linux, Alexey Alexandrov, Bill Wendling, Greg Thelen
On Mon, Oct 24, 2022 at 02:06:52PM -0700, Nick Desaulniers wrote:
> Dear stable kernel maintainers,
> Our production kernel team and ChromeOS kernel teams are reporting
> that they are unable to symbolize addresses of symbols defined in
> assembly sources due to a regression I caused with
> commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
I think you mean commit b8a9092330da ("Kbuild: do not emit debug info
for assembly with LLVM_IAS=1"), which is what you blamed in
32ef9e5054ec? a66049e2cf0e does not appear to have any affect on this
problem?
> I fixed this upstream with
> commit 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
> but I think this is infeasible to backport through to 4.19.y.
>
> Do the attached branch-specific variants look acceptable?
I think the stable specific versions should be okay. We should be able
to deal with any fallout that happens if there is any.
Acked-by: Nathan Chancellor <nathan@kernel.org>
Cheers,
Nathan
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
2022-10-25 16:53 ` Nathan Chancellor
@ 2022-10-26 17:50 ` Nick Desaulniers
0 siblings, 0 replies; 4+ messages in thread
From: Nick Desaulniers @ 2022-10-26 17:50 UTC (permalink / raw)
To: Nathan Chancellor
Cc: Greg KH, Sasha Levin, Masahiro Yamada, # 3.4.x,
clang-built-linux, Alexey Alexandrov, Bill Wendling, Greg Thelen
On Tue, Oct 25, 2022 at 9:53 AM Nathan Chancellor <nathan@kernel.org> wrote:
>
> On Mon, Oct 24, 2022 at 02:06:52PM -0700, Nick Desaulniers wrote:
> > Dear stable kernel maintainers,
> > Our production kernel team and ChromeOS kernel teams are reporting
> > that they are unable to symbolize addresses of symbols defined in
> > assembly sources due to a regression I caused with
> > commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
>
> I think you mean commit b8a9092330da ("Kbuild: do not emit debug info
> for assembly with LLVM_IAS=1"), which is what you blamed in
> 32ef9e5054ec? a66049e2cf0e does not appear to have any affect on this
> problem?
Ah right, grabbed the wrong sha from the commit messages.
Not an issue in the patches, just the initial email I sent. Should
still be good for stable maintainers to pick up.
>
> > I fixed this upstream with
> > commit 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
> > but I think this is infeasible to backport through to 4.19.y.
> >
> > Do the attached branch-specific variants look acceptable?
>
> I think the stable specific versions should be okay. We should be able
> to deal with any fallout that happens if there is any.
>
> Acked-by: Nathan Chancellor <nathan@kernel.org>
Thanks for triple checking!
--
Thanks,
~Nick Desaulniers
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
2022-10-24 21:06 backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files") Nick Desaulniers
2022-10-25 16:53 ` Nathan Chancellor
@ 2022-10-27 10:33 ` Greg KH
1 sibling, 0 replies; 4+ messages in thread
From: Greg KH @ 2022-10-27 10:33 UTC (permalink / raw)
To: Nick Desaulniers
Cc: Sasha Levin, Masahiro Yamada, # 3.4.x, Nathan Chancellor,
clang-built-linux, Alexey Alexandrov, Bill Wendling, Greg Thelen
On Mon, Oct 24, 2022 at 02:06:52PM -0700, Nick Desaulniers wrote:
> Dear stable kernel maintainers,
> Our production kernel team and ChromeOS kernel teams are reporting
> that they are unable to symbolize addresses of symbols defined in
> assembly sources due to a regression I caused with
> commit a66049e2cf0e ("Kbuild: make DWARF version a choice")
> I fixed this upstream with
> commit 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files")
> but I think this is infeasible to backport through to 4.19.y.
>
> Do the attached branch-specific variants look acceptable?
All now queued up, thanks.
greg k-h
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2022-10-27 10:34 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-24 21:06 backports of 32ef9e5054ec ("Makefile.debug: re-enable debug info for .S files") Nick Desaulniers
2022-10-25 16:53 ` Nathan Chancellor
2022-10-26 17:50 ` Nick Desaulniers
2022-10-27 10:33 ` Greg KH
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).