Hi, This patch fixes a potential data corruption in the index table of real files. demux_real.c:add_index_item() tries to grow the index table on demand if it exceeds current dimension. It grows it by halv the current size. However on first call to add_index_item, index_malloc_size is 0 even thou the index_table_size can be larger since it has been allocated at another point where index_malloc_size hasn't been set. This leads to setting index_malloc_size to 2048. If now index_table_size is larger 2048 already, realloc is called with a size which is smalled than current, and last part of that index table is lost. Regards elupus
Joakim Plate writes:
This patch fixes a potential data corruption in the index table of real files.
Do you have sample files? Diego
Hi, On Tuesday 01 March 2005 11:31, Diego Biurrun wrote:
Do you have sample files?
I reported the problem before and Reimar D?ffinger(?) give me a patch which will fix it. Somehow that patch seems have not get into CVS yet. Anyway it's hard to upload a sample, You need a complete file to use the index chunk in the rmvb file, and you need quite a large index chunk to crash mplayer, which means the file will be quite big. -- Best Regards, LR
Hi, On Tue, Mar 01, 2005 at 11:04:19PM +0800, Lu Ran wrote:
On Tuesday 01 March 2005 11:31, Diego Biurrun wrote:
Do you have sample files?
I reported the problem before and Reimar D?ffinger(?) give me a patch which will fix it. Somehow that patch seems have not get into CVS yet. Anyway it's
I remember something like this, but I can't find where this was discussed... Can you give me a hint? Anyway, I am quite certain that this patch is not from me (not my coding style), I actually think I suggested a slightly different patch. Greetings, Reimar Döffinger
Hi, On Wednesday 02 March 2005 18:21, Reimar Döffinger wrote:
I remember something like this, but I can't find where this was discussed... Can you give me a hint? Anyway, I am quite certain that this patch is not from me (not my coding style), I actually think I suggested a slightly different patch.
http://mplayerhq.hu/pipermail/mplayer-dev-eng/2005-January/032582.html -- Best Regards, LR
Hi, On Thu, Mar 03, 2005 at 01:14:13PM +0800, ?????? wrote:
On Wednesday 02 March 2005 18:21, Reimar Döffinger wrote:
I remember something like this, but I can't find where this was discussed... Can you give me a hint? Anyway, I am quite certain that this patch is not from me (not my coding style), I actually think I suggested a slightly different patch.
http://mplayerhq.hu/pipermail/mplayer-dev-eng/2005-January/032582.html
Thanks... For some reason I had marked it as applied, that's why I couldn't find it. I applied this variant as the problem seems to appear when the chunk is in one ig piece - the code with the while loop would allocate unneccesarily much memory. Greetings, Reimar Döffinger
participants (5)
-
Diego Biurrun -
Joakim Plate -
Lu Ran -
Reimar Döffinger -
陆然