From mboxrd@z Thu Jan  1 00:00:00 1970
References: <20210327095417.2289913-1-rpm@xenomai.org>
 <20210327095417.2289913-3-rpm@xenomai.org>
 <12f269c4-cd8f-3235-60a9-6f5029fe1529@siemens.com>
From: Philippe Gerum <rpm@xenomai.org>
Subject: Re: [PATCH 2/8] cobalt/mutex: Bring back ia32 support for
 mutex_timedwait
In-reply-to: <12f269c4-cd8f-3235-60a9-6f5029fe1529@siemens.com>
Date: Wed, 07 Apr 2021 18:57:07 +0200
Message-ID: <87v98ynicc.fsf@xenomai.org>
MIME-Version: 1.0
Content-Type: text/plain
List-Id: Discussions about the Xenomai project <xenomai.xenomai.org>
List-Unsubscribe: <https://xenomai.org/mailman/options/xenomai>,
 <mailto:xenomai-request@xenomai.org?subject=unsubscribe>
List-Archive: <http://xenomai.org/pipermail/xenomai/>
List-Post: <mailto:xenomai@xenomai.org>
List-Help: <mailto:xenomai-request@xenomai.org?subject=help>
List-Subscribe: <https://xenomai.org/mailman/listinfo/xenomai>,
 <mailto:xenomai-request@xenomai.org?subject=subscribe>
To: Jan Kiszka <jan.kiszka@siemens.com>
Cc: xenomai@xenomai.org, Florian Bezdeka <florian.bezdeka@siemens.com>


Jan Kiszka <jan.kiszka@siemens.com> writes:

> On 27.03.21 10:54, Philippe Gerum wrote:
>> From: Florian Bezdeka <florian.bezdeka@siemens.com>
>> 
>> The helper used for copying the timeout values (=mutex_fetch_timeout())
>> was always copying sizeof(struct timespec64) from user to kernel space.
>> For applications with time_t being 4 bytes only (like for native 32 bit
>> applications) that is simply too much.
>> 
>> Signed-off-by: Florian Bezdeka <florian.bezdeka@siemens.com>
>> ---
>>  kernel/cobalt/posix/mutex.c | 3 +--
>>  1 file changed, 1 insertion(+), 2 deletions(-)
>> 
>> diff --git a/kernel/cobalt/posix/mutex.c b/kernel/cobalt/posix/mutex.c
>> index d43a747b3..70fe7960a 100644
>> --- a/kernel/cobalt/posix/mutex.c
>> +++ b/kernel/cobalt/posix/mutex.c
>> @@ -349,8 +349,7 @@ COBALT_SYSCALL(mutex_lock, primary,
>>  static inline int mutex_fetch_timeout(struct timespec64 *ts,
>>  				      const void __user *u_ts)
>>  {
>> -	return u_ts == NULL ? -EFAULT :
>> -		cobalt_copy_from_user(ts, u_ts, sizeof(*ts));
>> +	return u_ts == NULL ? -EFAULT : cobalt_get_u_timespec(ts, u_ts);
>>  }
>>  
>>  COBALT_SYSCALL(mutex_timedlock, primary,
>> 
>
> Also here: Can we please avoid introducing a regression first and then
> fixing it in the same series? To my understanding, this belongs into
> patch 1, right?
>
> Same for patch 3 and 8 likely.
>
> Philippe, you can preserve Florian's fix-up credit simply in the commit
> message, if that was the concern.

That was the concern.

-- 
Philippe.