On Tue, Jun 03, 2008 at 07:57:00PM -0700, Matthew D'Asaro wrote:
Instead I set wm->info.lBufOffset to wm->info.lDiskOffset because the buffer is flushed at the beginning of a call to mmioSetBuffer so wm->info.lDiskOffset is synchronized with wm->info.lBufOffset and so wm->info.lDiskOffset contains the correct current file position. The test for mmio still passes with this patch.
wouldn't that indicate, that the current patch does not cover that case? can you add a test for it?