[OE-core] [daisy][PATCH 1/1] systemd: do not use alloca() function in case of uclibc

ChenQi Qi.Chen at windriver.com
Tue Jun 3 09:20:35 UTC 2014


On 06/03/2014 05:06 PM, Richard Purdie wrote:
> On Tue, 2014-06-03 at 15:42 +0800, Chen Qi wrote:
>> The alloca() function allocates space in the stack frame of the caller,
>> so using alloca(new_size - old_size) would possibly crash the stack,
>> causing a segment fault error.
>>
>> This patch fixes the above problem by avoiding using this function in
>> journal-file.c.
>>
>> [YOCTO #6201]
>>
>> Signed-off-by: Chen Qi <Qi.Chen at windriver.com>
>> ---
>>   .../0001-journal-file.c-do-not-use-alloca.patch    |   54 ++++++++++++++++++++
>>   meta/recipes-core/systemd/systemd_211.bb           |    1 +
>>   2 files changed, 55 insertions(+)
>>   create mode 100644 meta/recipes-core/systemd/systemd/0001-journal-file.c-do-not-use-alloca.patch
>>
>> diff --git a/meta/recipes-core/systemd/systemd/0001-journal-file.c-do-not-use-alloca.patch b/meta/recipes-core/systemd/systemd/0001-journal-file.c-do-not-use-alloca.patch
>> new file mode 100644
>> index 0000000..a638d58
>> --- /dev/null
>> +++ b/meta/recipes-core/systemd/systemd/0001-journal-file.c-do-not-use-alloca.patch
>> @@ -0,0 +1,54 @@
>> +Upstream-Status: Inappropriate [oe specific]
> >From the description, this sounds like an allocation error which can
> happen *anywhere* and is a problem that should be addressed upstream.
>
> This Upstream-Status field is therefore completely bogus. Its not
> inappropriate or oe specific. If you still believe it is, I'd like to
> hear more explanation.
>
> The abuses of this field are starting to really annoy me since this
> keeps happening.
>
> Cheers,
>
> Richard
>
>
>
>
>

Hi Richard,

The use of alloca() was introduced by an oe-specific patch from Khem Raj.

The patch is 
meta/recipes-core/systemd/systemd/systemd-pam-fix-fallocate.patch.
The upstream status of the above patch is as following.
        Upstream-Status: Denied [no desire for uclibc support]

That's why I use 'Inappropriate [oe specific]' in the Upstream-Status 
field of my patch.

And I just realized I forgot to also patch the journald-kmsg.c file. 
I'll send out a V2.

Best Regards,
Chen Qi



More information about the Openembedded-core mailing list