[PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[Qu Wenruo]

Allowing error injection for btrfs_check_leaf_full() and
btrfs_check_node() is useful to test the failure path of btrfs write
time tree check.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 fs/btrfs/tree-checker.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/fs/btrfs/tree-checker.c b/fs/btrfs/tree-checker.c
index b9dd2a6e133f..9c22a1fe38c1 100644
--- a/fs/btrfs/tree-checker.c
+++ b/fs/btrfs/tree-checker.c
@@ -15,6 +15,7 @@
  * carefully reviewed otherwise so it does not prevent mount of valid images.
  */
 
+#include <linux/error-injection.h>
 #include "ctree.h"
 #include "tree-checker.h"
 #include "disk-io.h"
@@ -942,6 +943,7 @@ int btrfs_check_leaf_full(struct extent_buffer *leaf)
 {
 	return check_leaf(leaf, true);
 }
+ALLOW_ERROR_INJECTION(btrfs_check_leaf_full, ERRNO);
 
 int btrfs_check_leaf_relaxed(struct extent_buffer *leaf)
 {
@@ -1005,3 +1007,4 @@ int btrfs_check_node(struct extent_buffer *node)
 out:
 	return ret;
 }
+ALLOW_ERROR_INJECTION(btrfs_check_node, ERRNO);
-- 
2.21.0

Re: [PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[David Sterba]

On Wed, Apr 24, 2019 at 03:22:53PM +0800, Qu Wenruo wrote:
> Allowing error injection for btrfs_check_leaf_full() and
> btrfs_check_node() is useful to test the failure path of btrfs write
> time tree check.
> 
> Signed-off-by: Qu Wenruo <wqu@suse.com>

Added to misc-next, thanks.

Re: [PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[David Sterba]

On Wed, Apr 24, 2019 at 03:22:53PM +0800, Qu Wenruo wrote:
> Allowing error injection for btrfs_check_leaf_full() and
> btrfs_check_node() is useful to test the failure path of btrfs write
> time tree check.
> 
> Signed-off-by: Qu Wenruo <wqu@suse.com>

Does not compile:

  CC [M]  fs/btrfs/tests/extent-map-tests.o
In file included from fs/btrfs/tree-checker.c:18:
./include/linux/error-injection.h:15:15: error: unknown type name ‘bool’
 static inline bool within_error_injection_list(unsigned long addr)
               ^~~~
./include/linux/error-injection.h: In function ‘within_error_injection_list’:
./include/linux/error-injection.h:17:9: error: ‘false’ undeclared (first use in this function)
  return false;
         ^~~~~
./include/linux/error-injection.h:17:9: note: each undeclared identifier is reported only once for each function it appears in
make[1]: *** [scripts/Makefile.build:275: fs/btrfs/tree-checker.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:1737: fs/btrfs/] Error 2


gcc is 8.3.1

Re: [PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[Qu Wenruo]

[-- Attachment #1.1: Type: text/plain, Size: 1893 bytes --]



On 2019/4/25 上午1:45, David Sterba wrote:
> On Wed, Apr 24, 2019 at 03:22:53PM +0800, Qu Wenruo wrote:
>> Allowing error injection for btrfs_check_leaf_full() and
>> btrfs_check_node() is useful to test the failure path of btrfs write
>> time tree check.
>>
>> Signed-off-by: Qu Wenruo <wqu@suse.com>
> 
> Does not compile:
> 
>   CC [M]  fs/btrfs/tests/extent-map-tests.o
> In file included from fs/btrfs/tree-checker.c:18:
> ./include/linux/error-injection.h:15:15: error: unknown type name ‘bool’
>  static inline bool within_error_injection_list(unsigned long addr)
>                ^~~~
> ./include/linux/error-injection.h: In function ‘within_error_injection_list’:
> ./include/linux/error-injection.h:17:9: error: ‘false’ undeclared (first use in this function)
>   return false;
>          ^~~~~
> ./include/linux/error-injection.h:17:9: note: each undeclared identifier is reported only once for each function it appears in
> make[1]: *** [scripts/Makefile.build:275: fs/btrfs/tree-checker.o] Error 1
> make[1]: *** Waiting for unfinished jobs....
> make: *** [Makefile:1737: fs/btrfs/] Error 2

Compiles without problem here using 8.2.1 gcc.

...
  CC [M]  fs/btrfs/check-integrity.o
  CC [M]  fs/btrfs/ref-verify.o
  CC [M]  fs/btrfs/tests/free-space-tests.o
  CC [M]  fs/btrfs/tests/extent-buffer-tests.o
  CC [M]  fs/btrfs/tests/btrfs-tests.o
  CC [M]  fs/btrfs/tests/extent-io-tests.o
  CC [M]  fs/btrfs/tests/inode-tests.o
  CC [M]  fs/btrfs/tests/qgroup-tests.o
  CC [M]  fs/btrfs/tests/free-space-tree-tests.o
  CC [M]  fs/btrfs/tests/extent-map-tests.o
  LD [M]  fs/btrfs/btrfs.o
  Building modules, stage 2.
  MODPOST 1 modules
  CC      fs/btrfs/btrfs.mod.o
  LD [M]  fs/btrfs/btrfs.ko

> 
> 
> gcc is 8.3.1
> 

Upgrading to gcc 8.3.0, still passes.

Is that a regression in gcc 8.3.1?

Thanks,
Qu


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[David Sterba]

On Thu, Apr 25, 2019 at 08:30:38AM +0800, Qu Wenruo wrote:
> 
> 
> On 2019/4/25 上午1:45, David Sterba wrote:
> > On Wed, Apr 24, 2019 at 03:22:53PM +0800, Qu Wenruo wrote:
> >> Allowing error injection for btrfs_check_leaf_full() and
> >> btrfs_check_node() is useful to test the failure path of btrfs write
> >> time tree check.
> >>
> >> Signed-off-by: Qu Wenruo <wqu@suse.com>
> > 
> > Does not compile:
> > 
> >   CC [M]  fs/btrfs/tests/extent-map-tests.o
> > In file included from fs/btrfs/tree-checker.c:18:
> > ./include/linux/error-injection.h:15:15: error: unknown type name ‘bool’
> >  static inline bool within_error_injection_list(unsigned long addr)
> >                ^~~~
> > ./include/linux/error-injection.h: In function ‘within_error_injection_list’:
> > ./include/linux/error-injection.h:17:9: error: ‘false’ undeclared (first use in this function)
> >   return false;
> >          ^~~~~
> > ./include/linux/error-injection.h:17:9: note: each undeclared identifier is reported only once for each function it appears in
> > make[1]: *** [scripts/Makefile.build:275: fs/btrfs/tree-checker.o] Error 1
> > make[1]: *** Waiting for unfinished jobs....
> > make: *** [Makefile:1737: fs/btrfs/] Error 2
> 
> Compiles without problem here using 8.2.1 gcc.
> 
> ...
>   CC [M]  fs/btrfs/check-integrity.o
>   CC [M]  fs/btrfs/ref-verify.o
>   CC [M]  fs/btrfs/tests/free-space-tests.o
>   CC [M]  fs/btrfs/tests/extent-buffer-tests.o
>   CC [M]  fs/btrfs/tests/btrfs-tests.o
>   CC [M]  fs/btrfs/tests/extent-io-tests.o
>   CC [M]  fs/btrfs/tests/inode-tests.o
>   CC [M]  fs/btrfs/tests/qgroup-tests.o
>   CC [M]  fs/btrfs/tests/free-space-tree-tests.o
>   CC [M]  fs/btrfs/tests/extent-map-tests.o
>   LD [M]  fs/btrfs/btrfs.o
>   Building modules, stage 2.
>   MODPOST 1 modules
>   CC      fs/btrfs/btrfs.mod.o
>   LD [M]  fs/btrfs/btrfs.ko
> 
> > 
> > 
> > gcc is 8.3.1
> > 
> 
> Upgrading to gcc 8.3.0, still passes.
> 
> Is that a regression in gcc 8.3.1?

I don't think it's caused by compiler but config options. And the build
bot sees the compilation failure too, so it's not specific to my setup:

https://lists.01.org/pipermail/kbuild-all/2019-April/060303.html

Re: [PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[Qu Wenruo]

[-- Attachment #1.1: Type: text/plain, Size: 2525 bytes --]



On 2019/4/25 下午10:13, David Sterba wrote:
> On Thu, Apr 25, 2019 at 08:30:38AM +0800, Qu Wenruo wrote:
>>
>>
>> On 2019/4/25 上午1:45, David Sterba wrote:
>>> On Wed, Apr 24, 2019 at 03:22:53PM +0800, Qu Wenruo wrote:
>>>> Allowing error injection for btrfs_check_leaf_full() and
>>>> btrfs_check_node() is useful to test the failure path of btrfs write
>>>> time tree check.
>>>>
>>>> Signed-off-by: Qu Wenruo <wqu@suse.com>
>>>
>>> Does not compile:
>>>
>>>   CC [M]  fs/btrfs/tests/extent-map-tests.o
>>> In file included from fs/btrfs/tree-checker.c:18:
>>> ./include/linux/error-injection.h:15:15: error: unknown type name ‘bool’
>>>  static inline bool within_error_injection_list(unsigned long addr)
>>>                ^~~~
>>> ./include/linux/error-injection.h: In function ‘within_error_injection_list’:
>>> ./include/linux/error-injection.h:17:9: error: ‘false’ undeclared (first use in this function)
>>>   return false;
>>>          ^~~~~
>>> ./include/linux/error-injection.h:17:9: note: each undeclared identifier is reported only once for each function it appears in
>>> make[1]: *** [scripts/Makefile.build:275: fs/btrfs/tree-checker.o] Error 1
>>> make[1]: *** Waiting for unfinished jobs....
>>> make: *** [Makefile:1737: fs/btrfs/] Error 2
>>
>> Compiles without problem here using 8.2.1 gcc.
>>
>> ...
>>   CC [M]  fs/btrfs/check-integrity.o
>>   CC [M]  fs/btrfs/ref-verify.o
>>   CC [M]  fs/btrfs/tests/free-space-tests.o
>>   CC [M]  fs/btrfs/tests/extent-buffer-tests.o
>>   CC [M]  fs/btrfs/tests/btrfs-tests.o
>>   CC [M]  fs/btrfs/tests/extent-io-tests.o
>>   CC [M]  fs/btrfs/tests/inode-tests.o
>>   CC [M]  fs/btrfs/tests/qgroup-tests.o
>>   CC [M]  fs/btrfs/tests/free-space-tree-tests.o
>>   CC [M]  fs/btrfs/tests/extent-map-tests.o
>>   LD [M]  fs/btrfs/btrfs.o
>>   Building modules, stage 2.
>>   MODPOST 1 modules
>>   CC      fs/btrfs/btrfs.mod.o
>>   LD [M]  fs/btrfs/btrfs.ko
>>
>>>
>>>
>>> gcc is 8.3.1
>>>
>>
>> Upgrading to gcc 8.3.0, still passes.
>>
>> Is that a regression in gcc 8.3.1?
> 
> I don't think it's caused by compiler but config options. And the build
> bot sees the compilation failure too, so it's not specific to my setup:
> 
> https://lists.01.org/pipermail/kbuild-all/2019-April/060303.html
> 

Got it reproduced, not sure why the config could affect this problem.

But the fix is always straightforward, include types for bool and stddef
for true/false.

Thanks,
Qu



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] btrfs: tree-checker: Allow error injection for tree-checker

[David Sterba]

On Thu, Apr 25, 2019 at 10:22:30PM +0800, Qu Wenruo wrote:
> 
> 
> On 2019/4/25 下午10:13, David Sterba wrote:
> > On Thu, Apr 25, 2019 at 08:30:38AM +0800, Qu Wenruo wrote:
> >>
> >>
> >> On 2019/4/25 上午1:45, David Sterba wrote:
> >>> On Wed, Apr 24, 2019 at 03:22:53PM +0800, Qu Wenruo wrote:
> >>>> Allowing error injection for btrfs_check_leaf_full() and
> >>>> btrfs_check_node() is useful to test the failure path of btrfs write
> >>>> time tree check.
> >>>>
> >>>> Signed-off-by: Qu Wenruo <wqu@suse.com>
> >>>
> >>> Does not compile:
> >>>
> >>>   CC [M]  fs/btrfs/tests/extent-map-tests.o
> >>> In file included from fs/btrfs/tree-checker.c:18:
> >>> ./include/linux/error-injection.h:15:15: error: unknown type name ‘bool’
> >>>  static inline bool within_error_injection_list(unsigned long addr)
> >>>                ^~~~
> >>> ./include/linux/error-injection.h: In function ‘within_error_injection_list’:
> >>> ./include/linux/error-injection.h:17:9: error: ‘false’ undeclared (first use in this function)
> >>>   return false;
> >>>          ^~~~~
> >>> ./include/linux/error-injection.h:17:9: note: each undeclared identifier is reported only once for each function it appears in
> >>> make[1]: *** [scripts/Makefile.build:275: fs/btrfs/tree-checker.o] Error 1
> >>> make[1]: *** Waiting for unfinished jobs....
> >>> make: *** [Makefile:1737: fs/btrfs/] Error 2
> >>
> >> Compiles without problem here using 8.2.1 gcc.
> >>
> >> ...
> >>   CC [M]  fs/btrfs/check-integrity.o
> >>   CC [M]  fs/btrfs/ref-verify.o
> >>   CC [M]  fs/btrfs/tests/free-space-tests.o
> >>   CC [M]  fs/btrfs/tests/extent-buffer-tests.o
> >>   CC [M]  fs/btrfs/tests/btrfs-tests.o
> >>   CC [M]  fs/btrfs/tests/extent-io-tests.o
> >>   CC [M]  fs/btrfs/tests/inode-tests.o
> >>   CC [M]  fs/btrfs/tests/qgroup-tests.o
> >>   CC [M]  fs/btrfs/tests/free-space-tree-tests.o
> >>   CC [M]  fs/btrfs/tests/extent-map-tests.o
> >>   LD [M]  fs/btrfs/btrfs.o
> >>   Building modules, stage 2.
> >>   MODPOST 1 modules
> >>   CC      fs/btrfs/btrfs.mod.o
> >>   LD [M]  fs/btrfs/btrfs.ko
> >>
> >>>
> >>>
> >>> gcc is 8.3.1
> >>>
> >>
> >> Upgrading to gcc 8.3.0, still passes.
> >>
> >> Is that a regression in gcc 8.3.1?
> > 
> > I don't think it's caused by compiler but config options. And the build
> > bot sees the compilation failure too, so it's not specific to my setup:
> > 
> > https://lists.01.org/pipermail/kbuild-all/2019-April/060303.html
> > 
> 
> Got it reproduced, not sure why the config could affect this problem.
> 
> But the fix is always straightforward, include types for bool and stddef
> for true/false.

As this is only a minor compilation issue, I'll add the includes so we
don't have to wait on the proper fix to land and you can use the
injection.