From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:57213) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fJd0j-0002DO-13 for qemu-devel@nongnu.org; Fri, 18 May 2018 06:53:22 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1fJd0f-0007Ad-3E for qemu-devel@nongnu.org; Fri, 18 May 2018 06:53:21 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:47766 helo=mx1.redhat.com) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1fJd0e-0007A6-TY for qemu-devel@nongnu.org; Fri, 18 May 2018 06:53:17 -0400 Received: from smtp.corp.redhat.com (int-mx04.intmail.prod.int.rdu2.redhat.com [10.11.54.4]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 45A0A401EF14 for ; Fri, 18 May 2018 10:53:16 +0000 (UTC) Date: Fri, 18 May 2018 18:53:11 +0800 From: Peter Xu Message-ID: <20180518105311.GP2569@xz-mi> References: <20180509041734.14135-1-peterx@redhat.com> <20180509041734.14135-5-peterx@redhat.com> <87tvr6l9ix.fsf@dusky.pond.sub.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <87tvr6l9ix.fsf@dusky.pond.sub.org> Subject: Re: [Qemu-devel] [PATCH v5 4/4] monitor: add lock to protect mon_fdsets List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Markus Armbruster Cc: qemu-devel@nongnu.org, "Dr . David Alan Gilbert" , Stefan Hajnoczi , =?utf-8?Q?Marc-Andr=C3=A9?= Lureau On Thu, May 17, 2018 at 03:03:02PM +0200, Markus Armbruster wrote: [...] > > @@ -2502,7 +2525,9 @@ int monitor_fdset_get_fd(int64_t fdset_id, int flags) > > MonFdset *mon_fdset; > > MonFdsetFd *mon_fdset_fd; > > int mon_fd_flags; > > + int ret = -1; > > Suggest not to initialize ret, and instead ret = -1 on both failure > paths. [1] But there is a third hidden failure path that we failed to find the fd specified? In that case we still need that initial value. But I didn't really notice that this function is returning error with -1 paired with errno. So instead of set -1 here I may need to initialize it to -ENOENT, and I can convert it back to errno when return. Please see below. > > > > > + qemu_mutex_lock(&mon_fdsets_lock); > > QLIST_FOREACH(mon_fdset, &mon_fdsets, next) { > > if (mon_fdset->id != fdset_id) { > > continue; > > @@ -2510,49 +2535,62 @@ int monitor_fdset_get_fd(int64_t fdset_id, int flags) > > QLIST_FOREACH(mon_fdset_fd, &mon_fdset->fds, next) { > > mon_fd_flags = fcntl(mon_fdset_fd->fd, F_GETFL); > > if (mon_fd_flags == -1) { > > - return -1; > > + goto out; > > Preexisting: we fail without setting errno. Smells buggy. Indeed. Here I possibly need to set "ret = -errno" since at [2] below the errno might be polluted by the mutex unlocking operation. > > Can we avoid setting errno and return a negative errno code instead? Yes that'll be nice, but it's getting out of the scope of this patchset. So I'll try to avoid touching that. I mean qemu_open() and its callers. > > > } > > > > if ((flags & O_ACCMODE) == (mon_fd_flags & O_ACCMODE)) { > > - return mon_fdset_fd->fd; > > + ret = mon_fdset_fd->fd; > > + goto out; > > } > > } > > errno = EACCES; > > - return -1; > > + break; > > } > > -#endif > > - > > +out: > > + qemu_mutex_unlock(&mon_fdsets_lock); [2] > > + return ret; So in my next post I'll make sure I return -1 when error happens, and errno should contain the correct (positive) error. > > +#else > > #ifndef _WIN32 ... #endif becomes #ifndef _WIN32 ... #else ... #endif. > I hate negative conditionals with else. Mind to swap? Sure I can. > > > errno = ENOENT; > > return -1; > > +#endif > > } > > > > int monitor_fdset_dup_fd_add(int64_t fdset_id, int dup_fd) > > { > > MonFdset *mon_fdset; > > MonFdsetFd *mon_fdset_fd_dup; > > + int ret = -1; > > Dead initializer, please remove. IMHO it's the same as above [1], so we still need that, right? > > > > > + qemu_mutex_lock(&mon_fdsets_lock); > > QLIST_FOREACH(mon_fdset, &mon_fdsets, next) { > > if (mon_fdset->id != fdset_id) { > > continue; > > } > > QLIST_FOREACH(mon_fdset_fd_dup, &mon_fdset->dup_fds, next) { > > if (mon_fdset_fd_dup->fd == dup_fd) { > > - return -1; > > + ret = -1; > > + goto out; > > } > > } > > mon_fdset_fd_dup = g_malloc0(sizeof(*mon_fdset_fd_dup)); > > mon_fdset_fd_dup->fd = dup_fd; > > QLIST_INSERT_HEAD(&mon_fdset->dup_fds, mon_fdset_fd_dup, next); > > - return 0; > > + ret = 0; > > + break; > > } > > - return -1; > > + > > +out: > > + qemu_mutex_unlock(&mon_fdsets_lock); > > + return ret; > > } > > > > static int monitor_fdset_dup_fd_find_remove(int dup_fd, bool remove) > > { > > MonFdset *mon_fdset; > > MonFdsetFd *mon_fdset_fd_dup; > > + int ret = -1; > > Likewise. Same as [1]? Thanks, -- Peter Xu