Messages in this thread | | | Date | Tue, 25 Apr 2023 20:13:27 +0900 | From | asmadeus@codewrec ... | Subject | Re: [PATCH v5] fs/9p: remove writeback fid and fix per-file modes |
| |
Christophe JAILLET wrote on Tue, Apr 25, 2023 at 09:11:01AM +0200: > This code looks strange. > P9_OWRITE is 0x01, so !P9_OWRITE is 0. > So the code is equivalent to "p9_omode = P9_ORDWR;" > > Is it what is expexted? > > Maybe > p9_omode = (p9_omode & ~P9_OWRITE) | P9_ORDWR; > ?
Since we're discussing tooling, sparse caught this one: fs/9p/vfs_inode.c:826:38: warning: dubious: x & !y fs/9p/vfs_inode_dotl.c:291:38: warning: dubious: x & !y
(runing with make `M=fs/9p W=1 C=2` ; unfortunately error code doesn't reflect a problem so that'll require inspecting output to automate...)
I've tried running scan-build in a very old-fashioned way (getting the gcc lines from make V=1 and re-running with scan-build) and it had some arguable warnings (setting `ret = 0` before it is overwritten again is considered a dead store), but it had some real problems as well so it might be worth fixing these just to reduce the clutter and run it all the time. I'll post a couple of patches tomorrow (unrelated to this)
-- Dominique
| |