Emmanuel Dreyfus <[email protected]> wrote: > Here is my analysis of the problem so far: the kernel is nice enough to > call puffs_vnop_fsync before puffs_vnop_rename, therefore the data > should be sync to the file and I should not see that bug.
Erratum: adding a backtrace on puffs_vnop_fsync tells me that that call is triggered by the unlink(2). The kernel makes no attempt to sync before rename. I tried to add a VOP_PUTPAGES in puffs_vnop_rename so that the dirty page got pushed to the file before the rename, but without any success. The VM system seems to consiser there are no dirty pages, although I did write a few bytes in the file. I have trouble to understand why. -- Emmanuel Dreyfus http://hcpnet.free.fr/pubz [email protected]
