Date: Sun, 7 Sep 2025 11:21:54 -0600 From: Alan Somers <asomers@freebsd.org> To: Konstantin Belousov <kostikbel@gmail.com> Cc: obiwac <obiwac@freebsd.org>, freebsd-current@freebsd.org, freebsd-fs@freebsd.org Subject: Re: Fuse issue with attr caching expecting write lock Message-ID: <CAOtMX2gr2uFBHbvHDgcwZiihYZ5tCNKxmz6=5bY2F=vfu7O5Ug@mail.gmail.com> In-Reply-To: <aLorc9rz87dibsQL@kib.kiev.ua> References: <CAN8-kNV6-H_YtmT2MeYkYOPcgcWrpe2K9jxqdb%2BjRzOJuk=-dA@mail.gmail.com> <aLorc9rz87dibsQL@kib.kiev.ua>
index | next in thread | previous in thread | raw e-mail
[-- Attachment #1 --] On Thu, Sep 4, 2025 at 6:15 PM Konstantin Belousov <kostikbel@gmail.com> wrote: > On Fri, Sep 05, 2025 at 01:36:39AM +0200, obiwac wrote: > > Hi! > > > > When trying to access a large file with sshfs (on fuse), I get a > > kernel panic because fuse_internal_cache_attrs is asserting a write > > lock when it is sometimes being called from fuse_internal_do_getattr, > > which only asserts a read lock. > > > > To me (i.e. someone who has never touched VFS or fuse code) this code > > looks wrong, and fuse_internal_do_getattr should be acquiring a write > > lock around its call to fuse_internal_cache_attrs, but clearly it has > > been working just fine for a while now as this code hasn't changed > > super recently so I could be missing something. > It is only the assertions that were enabled recently. > > Mark merged DEBUG_VFS_LOCKS option into INVARIANTS. > I'm not surprised to see bugs like this one, for the reason Kib stated. I suspect that there are many more, because DEBUG_VFS_LOCKS was historically rarely used. If you can come up with an easy reproduction case, please open a ticket on bugzilla and CC me. And BTW, my FreeBSD user name is "asomers", not "alan.somers". [-- Attachment #2 --] <div dir="ltr"><div dir="ltr"><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Sep 4, 2025 at 6:15 PM Konstantin Belousov <<a href="mailto:kostikbel@gmail.com" target="_blank">kostikbel@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On Fri, Sep 05, 2025 at 01:36:39AM +0200, obiwac wrote:<br> > Hi!<br> > <br> > When trying to access a large file with sshfs (on fuse), I get a<br> > kernel panic because fuse_internal_cache_attrs is asserting a write<br> > lock when it is sometimes being called from fuse_internal_do_getattr,<br> > which only asserts a read lock.<br> > <br> > To me (i.e. someone who has never touched VFS or fuse code) this code<br> > looks wrong, and fuse_internal_do_getattr should be acquiring a write<br> > lock around its call to fuse_internal_cache_attrs, but clearly it has<br> > been working just fine for a while now as this code hasn't changed<br> > super recently so I could be missing something.<br> It is only the assertions that were enabled recently.<br> <br> Mark merged DEBUG_VFS_LOCKS option into INVARIANTS.<br></blockquote><div><br></div><div> I'm not surprised to see bugs like this one, for the reason Kib stated. I suspect that there are many more, because DEBUG_VFS_LOCKS was historically rarely used. If you can come up with an easy reproduction case, please open a ticket on bugzilla and CC me. And BTW, my FreeBSD user name is "asomers", not "alan.somers".</div></div></div> </div>home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAOtMX2gr2uFBHbvHDgcwZiihYZ5tCNKxmz6=5bY2F=vfu7O5Ug>