"Vladimir V. Saveliev" wrote:
>
> ... you are hitting a problem known as "not perfect handling of race conditions
> occuring during writing pages to a sparse file".
>
> There are two fixes for this problem:
> first is "expanding truncate" (although I am not sure which kernel it is for)
> second changes reiserfs_get_block (patch against 2.4.8) to handle this race
>properly. It is
> probably less tested then the first one, it survives your mmap test for me though.
I tried the reiserfs_get_block patch on my embedded system and it
fixes the problem. I'll run more tests through it to check it the
stability.
Thank you very much for your prompt response! You guys are
great!
Wilson Ho
