Re: 2.6.25-rc7-git5: Reported regressions from 2.6.24

[Mark Lord <lkml@rtr.ca>]

Adrian Bunk wrote:
> On Mon, Mar 31, 2008 at 01:48:15PM -0400, Mark Lord wrote:
>> Rafael,
>>
>> Add this one to the list?
> 
> 
> That's not a regression from 2.6.24
..

2.6.24 does not flood my syslog with those messages.
2.6.25-rc* does.   Looks like a regression.

The original bug was a hidden regression in 2.6.23,
which nobody bothered to identify until now.

Just because some code "gets away with" a regression
for a kernel or two, doesn't mean that regression shouldn't be fixed.

Cheers



> 
> 
>> * * * * *
>>
>> Subject: + driver-core-fix-uevent-action-string-regression.patch added to -mm tree
>> From: akpm
>> To: mm-commits
>> Date: 28/03/08 07:51 PM
>>
>> The patch titled
>>     driver core: fix uevent action-string regression
>> has been added to the -mm tree.  Its filename is
>>     driver-core-fix-uevent-action-string-regression.patch
>> ...
>>
>> The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/
>>
>> ------------------------------------------------------
>> Subject: driver core: fix uevent action-string regression
>> From: Mark Lord <lkml@rtr.ca>
>>
>> On boot, syslog is flooded with "uevent: unsupported action-string;" messages.
>> ..
>> Mar 28 14:43:29 shrimp kernel: tty ptyqd: uevent: unsupported
>> action-string; this will be ignored in a future kernel version
>> Mar 28 14:43:29 shrimp kernel: tty ptyqe: uevent: unsupported
>> action-string; this will be ignored in a future kernel version
>> Mar 28 14:43:29 shrimp kernel: tty ptyqf: uevent: unsupported
>> action-string; this will be ignored in a future kernel version
>> Mar 28 14:43:29 shrimp kernel: tty ptyr0: uevent: unsupported
>> action-string; this will be ignored in a future kernel version
>>
>>
>> These messages are a regression compared with 2.6.24, which did not flood the
>> syslog with them.
>>
>> The actual underlying problem was introduced in 2.6.23, when somebody made the
>> string parsing no longer accept nul-terminated strings as a valid input to
>> store_uevent().
>>
>> Eg.  "add\0" was valid prior to 2.6.23, where the code regressed to require
>> "add" without the '\0'.
>>
>> This patch fixes the 2.6.23 / 2.6.24 regressions, by having the code once
>> again tolerate the trailing '\0', if present.
>>
>> According to GregKH, this mainly affects older Ubuntu systems, such as the one
>> I have here that requires this fix.
>>
>> Signed-off-by: Mark Lord <mlord@pobox.com>
>> Cc: Greg KH <greg@kroah.com>
>> Cc: Kay Sievers <kay.sievers@vrfy.org>
>> Cc: <stable@kernel.org>
>> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
>> ---
>>
>> lib/kobject_uevent.c |    2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff -puN lib/kobject_uevent.c~driver-core-fix-uevent-action-string-regression lib/kobject_uevent.c
>> --- a/lib/kobject_uevent.c~driver-core-fix-uevent-action-string-regression
>> +++ a/lib/kobject_uevent.c
>> @@ -59,7 +59,7 @@ int kobject_action_type(const char *buf,
>> 	enum kobject_action action;
>> 	int ret = -EINVAL;
>>
>> -	if (count && buf[count-1] == '\n')
>> +	if (count && (buf[count-1] == '\n' || buf[count-1] == '\0'))
>> 		count--;
>>
>> 	if (!count)
>> _
>>
>> Patches currently in -mm which might be from lkml@rtr.ca are
>>
>> driver-core-fix-uevent-action-string-regression.patch
> 
> cu
> Adrian
>