Skip to contentSkip to conversation

Crash in -next due to 'objtool: Improve detection of BUG() and other dead ends'

PATCH New Guenter Roeck 3 participants 6 messages Feb 27, 2017 – Feb 28, 2017

lore
Share & tools
lore
#0 lore raw

Hi,

my qemu tests for mips64 in -next fail as follows.

...
VFS: Mounted root (ext3 filesystem) on device 8:0. Freeing unused kernel memory: 304K
This architecture does not have kernel memory protection.

Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

There is no symbol traceback, and I don't see any other (obvious) error message. bisect points to commit 'objtool: Improve detection of BUG() and other dead ends'; reverting that patch fixes the problem. Bisect log is attached below.

The script used to run the test as well as the root file system is available at https://github.com/groeck/linux-build-test/tree/master/rootfs/mips64

A complete log is available at
http://kerneltests.org/builders/qemu-mips64-next/builds/592/steps/qemubuildcommand/logs/stdio

Guenter

---
# bad: [ed7b11e565c736828f0b793f596a4ca20efee747] Add linux-next specific files for 20170227
# good: [c470abd4fde40ea6a0846a2beab642a578c0b8cd] Linux 4.10
git bisect start 'HEAD' 'v4.10'
# good: [caa59428971d5ad81d19512365c9ba580d83268c] Merge tag 'staging-4.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
git bisect good caa59428971d5ad81d19512365c9ba580d83268c
# good: [d5500a074741b78b7f778b4ab3415d5ecdcda0a7] Merge tag 'usercopy-v4.11-rc1.fix' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux
git bisect good d5500a074741b78b7f778b4ab3415d5ecdcda0a7
# good: [5d8a00eee2ed2e548a5d21b0edf495f3f7bf8bb4] Merge tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux
git bisect good 5d8a00eee2ed2e548a5d21b0edf495f3f7bf8bb4
# good: [017bfb6c372fa793cdb09828597a464fa24e0ea7] Merge remote-tracking branch 'jfs/jfs-next'
git bisect good 017bfb6c372fa793cdb09828597a464fa24e0ea7
# bad: [e2a95a3fe9e9586df29e8d673dc8cdd36ed6158b] Merge remote-tracking branch 'edac/linux_next'
git bisect bad e2a95a3fe9e9586df29e8d673dc8cdd36ed6158b
# good: [736e441efdafdbf0bb32732d87e00b8ba7f75260] Merge remote-tracking branch 'thermal/next'
git bisect good 736e441efdafdbf0bb32732d87e00b8ba7f75260
# good: [5bd695c24b294b482d86d352a4a0c7ba2059dcff] Merge remote-tracking branch 'block/for-next'
git bisect good 5bd695c24b294b482d86d352a4a0c7ba2059dcff
# good: [690962e72babac64c8240071e106d2c3e8d08306] Merge remote-tracking branch 'spi/for-next'
git bisect good 690962e72babac64c8240071e106d2c3e8d08306
# bad: [462a53b1a2f984a8c8c0aade1c461ce9ea156d5b] Merge branch 'perf/urgent'
git bisect bad 462a53b1a2f984a8c8c0aade1c461ce9ea156d5b
# good: [bb963e16507ca7670f0bb47ccaada8874b2ba6a1] perf utils: Check verbose flag properly
git bisect good bb963e16507ca7670f0bb47ccaada8874b2ba6a1
# good: [7bbba0eb1af34694868d028b80475981f90e6bee] perf/core: Fix perf_event_enable_on_exec() timekeeping (again)
git bisect good 7bbba0eb1af34694868d028b80475981f90e6bee
# bad: [1db159781cc7df093085ccaddc022beaaa9dd990] Merge branch 'core/urgent'
git bisect bad 1db159781cc7df093085ccaddc022beaaa9dd990
# bad: [d1091c7fa3d52ebce4dd3f15d04155b3469b2f90] objtool: Improve detection of BUG() and other dead ends
git bisect bad d1091c7fa3d52ebce4dd3f15d04155b3469b2f90
# good: [9f0c18aec620bc9d82268b3cb937568dd07b43ff] objtool: Fix CONFIG_STACK_VALIDATION=y warning for out-of-tree modules git bisect good 9f0c18aec620bc9d82268b3cb937568dd07b43ff

# first bad commit: [d1091c7fa3d52ebce4dd3f15d04155b3469b2f90] objtool: Improve detection of BUG() and other dead ends
#1 lore raw
Quoted 17 lines · Guenter Roeck

Hi,

my qemu tests for mips64 in -next fail as follows.

...
VFS: Mounted root (ext3 filesystem) on device 8:0. Freeing unused kernel memory: 304K
This architecture does not have kernel memory protection.

Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

There is no symbol traceback, and I don't see any other (obvious) error message. bisect points to commit 'objtool: Improve detection of BUG() and other dead ends'; reverting that patch fixes the problem. Bisect log is attached below.

Quoted 5 lines

The script used to run the test as well as the root file system is available at https://github.com/groeck/linux-build-test/tree/master/rootfs/mips64

A complete log is available at
http://kerneltests.org/builders/qemu-mips64-next/builds/592/steps/qemubuildcommand/logs/stdio

Hi Guenter,

This failure doesn't make much sense. Objtool (and the reported bad patch) are x86-only and should have nothing to do with mips64. Would it be possible for you to double check the error?

Otherwise I could try to recreate. Do you have a recommendation for a mips cross-compiler? (preferably something that's already built and that might work on Fedora)

Quoted 33 lines

---
# bad: [ed7b11e565c736828f0b793f596a4ca20efee747] Add linux-next specific files for 20170227
# good: [c470abd4fde40ea6a0846a2beab642a578c0b8cd] Linux 4.10
git bisect start 'HEAD' 'v4.10'
# good: [caa59428971d5ad81d19512365c9ba580d83268c] Merge tag 'staging-4.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
git bisect good caa59428971d5ad81d19512365c9ba580d83268c
# good: [d5500a074741b78b7f778b4ab3415d5ecdcda0a7] Merge tag 'usercopy-v4.11-rc1.fix' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux
git bisect good d5500a074741b78b7f778b4ab3415d5ecdcda0a7
# good: [5d8a00eee2ed2e548a5d21b0edf495f3f7bf8bb4] Merge tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux
git bisect good 5d8a00eee2ed2e548a5d21b0edf495f3f7bf8bb4
# good: [017bfb6c372fa793cdb09828597a464fa24e0ea7] Merge remote-tracking branch 'jfs/jfs-next'
git bisect good 017bfb6c372fa793cdb09828597a464fa24e0ea7
# bad: [e2a95a3fe9e9586df29e8d673dc8cdd36ed6158b] Merge remote-tracking branch 'edac/linux_next'
git bisect bad e2a95a3fe9e9586df29e8d673dc8cdd36ed6158b
# good: [736e441efdafdbf0bb32732d87e00b8ba7f75260] Merge remote-tracking branch 'thermal/next'
git bisect good 736e441efdafdbf0bb32732d87e00b8ba7f75260
# good: [5bd695c24b294b482d86d352a4a0c7ba2059dcff] Merge remote-tracking branch 'block/for-next'
git bisect good 5bd695c24b294b482d86d352a4a0c7ba2059dcff
# good: [690962e72babac64c8240071e106d2c3e8d08306] Merge remote-tracking branch 'spi/for-next'
git bisect good 690962e72babac64c8240071e106d2c3e8d08306
# bad: [462a53b1a2f984a8c8c0aade1c461ce9ea156d5b] Merge branch 'perf/urgent'
git bisect bad 462a53b1a2f984a8c8c0aade1c461ce9ea156d5b
# good: [bb963e16507ca7670f0bb47ccaada8874b2ba6a1] perf utils: Check verbose flag properly
git bisect good bb963e16507ca7670f0bb47ccaada8874b2ba6a1
# good: [7bbba0eb1af34694868d028b80475981f90e6bee] perf/core: Fix perf_event_enable_on_exec() timekeeping (again)
git bisect good 7bbba0eb1af34694868d028b80475981f90e6bee
# bad: [1db159781cc7df093085ccaddc022beaaa9dd990] Merge branch 'core/urgent'
git bisect bad 1db159781cc7df093085ccaddc022beaaa9dd990
# bad: [d1091c7fa3d52ebce4dd3f15d04155b3469b2f90] objtool: Improve detection of BUG() and other dead ends
git bisect bad d1091c7fa3d52ebce4dd3f15d04155b3469b2f90
# good: [9f0c18aec620bc9d82268b3cb937568dd07b43ff] objtool: Fix CONFIG_STACK_VALIDATION=y warning for out-of-tree modules git bisect good 9f0c18aec620bc9d82268b3cb937568dd07b43ff

# first bad commit: [d1091c7fa3d52ebce4dd3f15d04155b3469b2f90] objtool: Improve detection of BUG() and other dead ends
Signature · 1 line
Josh
#2 lore raw

Hi Josh,

Quoted 21 lines · Josh Poimboeuf
Quoted 20 lines · Guenter Roeck

Hi,

my qemu tests for mips64 in -next fail as follows.

...
VFS: Mounted root (ext3 filesystem) on device 8:0. Freeing unused kernel memory: 304K
This architecture does not have kernel memory protection.

Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

There is no symbol traceback, and I don't see any other (obvious) error message. bisect points to commit 'objtool: Improve detection of BUG() and other dead ends'; reverting that patch fixes the problem. Bisect log is attached below.

The script used to run the test as well as the root file system is available at https://github.com/groeck/linux-build-test/tree/master/rootfs/mips64

Quoted 9 lines
Quoted 2 lines

A complete log is available at
http://kerneltests.org/builders/qemu-mips64-next/builds/592/steps/qemubuildcommand/logs/stdio

Hi Guenter,

This failure doesn't make much sense. Objtool (and the reported bad patch) are x86-only and should have nothing to do with mips64. Would it be possible for you to double check the error?

I already did; after all, bisect points to the patch, and reverting it fixes the problem. After looking into the patch, I concluded that the following might help.

-#define unreachable() annotate_unreachable(); __builtin_unreachable() +#define unreachable() do { annotate_unreachable(); __builtin_unreachable(); } while (0)

and it does ...

Quoted 4 lines

Otherwise I could try to recreate. Do you have a recommendation for a mips cross-compiler? (preferably something that's already built and that might work on Fedora)

I use gcc 4.9.0 from
https://www.kernel.org/pub/tools/crosstool/files/bin/x86_64/4.9.0/

Guenter

#3 lore raw
Quoted 40 lines · Guenter Roeck

Hi Josh,

Quoted 21 lines · Josh Poimboeuf
Quoted 20 lines · Guenter Roeck

Hi,

my qemu tests for mips64 in -next fail as follows.

...
VFS: Mounted root (ext3 filesystem) on device 8:0. Freeing unused kernel memory: 304K
This architecture does not have kernel memory protection.

Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000a

There is no symbol traceback, and I don't see any other (obvious) error message. bisect points to commit 'objtool: Improve detection of BUG() and other dead ends'; reverting that patch fixes the problem. Bisect log is attached below.

The script used to run the test as well as the root file system is available at https://github.com/groeck/linux-build-test/tree/master/rootfs/mips64

Quoted 9 lines
Quoted 2 lines

A complete log is available at
http://kerneltests.org/builders/qemu-mips64-next/builds/592/steps/qemubuildcommand/logs/stdio

Hi Guenter,

This failure doesn't make much sense. Objtool (and the reported bad patch) are x86-only and should have nothing to do with mips64. Would it be possible for you to double check the error?

I already did; after all, bisect points to the patch, and reverting it fixes the problem. After looking into the patch, I concluded that the following might help.

(quoted diff) +1 -1
-#define unreachable() annotate_unreachable(); __builtin_unreachable()
+#define unreachable() do { annotate_unreachable(); __builtin_unreachable(); } while (0)
Quoted 2 lines

and it does ...

Doh, of course. Thanks for pointing it out, I'll fix it.

Signature · 1 line
Josh
#4 lore raw

PATCH 1/1 objtool: enclose contents of unreachable() macro in a block

Guenter Roeck reported a boot failure in mips64. It was bisected to the following commit:

d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead ends")

The unreachable() macro was formerly only composed of a single statement. The above commit added a second statement, but neglected to enclose the statements in a block.

Fixes d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead ends")
---
 include/linux/compiler-gcc.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
include/linux/compiler-gcc.h +2 -1
index de47134..f457b520 100644
@@ -215,7 +215,8 @@
  * this in the preprocessor, but we can live with this because they're
  * unreleased.  Really, we need to have autoconf for the kernel.
  */
-#define unreachable() annotate_unreachable(); __builtin_unreachable()
+#define unreachable() \
+	do { annotate_unreachable(); __builtin_unreachable(); } while (0)
 
 /* Mark a function definition as prohibited from being cloned. */
 #define __noclone	__attribute__((__noclone__, __optimize__("no-tracer")))
Signature · 1 line
2.7.4
#5 lore raw

[tip:core/urgent] objtool: Enclose contents of unreachable() macro in a block

objtool: Enclose contents of unreachable() macro in a block
Commit-ID:  4e4636cf981b5b629fbfb78aa9f232e015f7d521
Gitweb:     http://git.kernel.org/tip/4e4636cf981b5b629fbfb78aa9f232e015f7d521
Author:     Josh Poimboeuf <[email protected]>
AuthorDate: Mon, 27 Feb 2017 22:21:16 -0600
Committer:  Ingo Molnar <[email protected]>
CommitDate: Tue, 28 Feb 2017 07:47:26 +0100

objtool: Enclose contents of unreachable() macro in a block

Guenter Roeck reported a boot failure in mips64. It was bisected to the following commit:

d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead ends")

The unreachable() macro was formerly only composed of a single statement. The above commit added a second statement, but neglected to enclose the statements in a block.

Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead ends")
---
 include/linux/compiler-gcc.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
include/linux/compiler-gcc.h +2 -1
index de47134..f457b52 100644
@@ -215,7 +215,8 @@
  * this in the preprocessor, but we can live with this because they're
  * unreleased.  Really, we need to have autoconf for the kernel.
  */
-#define unreachable() annotate_unreachable(); __builtin_unreachable()
+#define unreachable() \
+	do { annotate_unreachable(); __builtin_unreachable(); } while (0)
 
 /* Mark a function definition as prohibited from being cloned. */
 #define __noclone	__attribute__((__noclone__, __optimize__("no-tracer")))