Re: [PATCH RERESEND 00/11] splice(file<>pipe) I/O on file as-if O_NONBLOCK
On Thu, Dec 14, 2023 at 12:06:57PM -0700, Jens Axboe wrote: > On 12/14/23 11:44 AM, Ahelenia Ziemiańska wrote: > > This does that, effectively making splice(file -> pipe) > > request (and require) O_NONBLOCK on reads fron the file: > > this doesn't affect splicing from regular files and blockdevs, > > since they're always non-blocking > > (and requesting the stronger "no kernel sleep" IOCB_NOWAIT is non-sensical), > Not sure how you got the idea that regular files or block devices is > always non-blocking, this is certainly not true without IOCB_NOWAIT. > Without IOCB_NOWAIT, you can certainly be waiting for previous IO to > complete. Maybe "always non-blocking" is an abuse of the term, but the terminology is lost on me. By this I mean that O_NONBLOCK files/blockdevs have the same semantics as non-O_NONBLOCK files/blockdevs ‒ they may block for a bit while the I/O queue drains, but are guaranteed to complete within a relatively narrow bounded time; any contending writer/opener will be blocked for a short bit but will always wake up. This is in contrast to pipes/sockets/ttys/&c., which wait for a peer to send some data, and block until there is some; any contending writer/opener will be blocked potentially ad infinitum. Or, the way I see it, splice(socket -> pipe) can trivially be used to lock the pipe forever, whereas I don't think splice(regfile -> pipe) can, regardless of IOCB_NOWAIT, so the specific semantic IOCB_NOWAIT provides is immaterial here, so not specifying IOCB_NOWAIT in filemap_splice_read() provides semantics consistent to "file is read as-if it had O_NONBLOCK set". > > but always returns -EINVAL for ttys. > > Sockets behave as expected from O_NONBLOCK reads: > > splice if there's data available else -EAGAIN. > > > > This should all pretty much behave as-expected. > Should it? Seems like there's a very high risk of breaking existing use > cases here. If something wants to splice from a socket to a pipe and doesn't degrade to read/write if it gets EAGAIN then it will either retry and hotloop in the splice or error out, yeah. I don't think this is surmountable. > Have you at all looked into the approach of enabling splice to/from > _without_ holding the pipe lock? That, to me, would seem like a much > saner approach, with the caveat that I have not looked into that at all > so there may indeed be reasons why this is not feasible. IIUC Linus prepared a patch on security@ in (you're in To:) and an evolution of this is in https://lore.kernel.org/lkml/CAHk-=wgmld78uslu9a9nspxytm9s6c23ovdin2yja-d8_s0...@mail.gmail.com/t/#u (you're in Cc:) that does this. He summarises it below as > So while fixing your NULL pointer check should be trivial, I think > that first patch is actually fundamentally broken wrt pipe resizing, > and I see no really sane way to fix it. We could add a new lock just > for that, but I don't think it's worth it. and > But it is possible that we need to just bite the bullet and say > "copy_splice_read() needs to use a non-blocking kiocb for the IO". so that's what I did. If Linus, who drew up and maintained this code for ~30 years, didn't arrive at a satisfactory approach, I, after ~30 minutes, won't either. It would be very sane to both not change the semantic and fix the lock by just not locking but at the top of that thread Christian said > Splice would have to be > refactored to not rely on pipe_lock(). That's likely major work with a > good portion of regressions if the past is any indication. and clearly this is true, so lacking the ability and capability to do that and any reasonable other ideas (You could either limit splices to a proportion of the pipe size, but then just doing five splices gets you where we are rn (or, as Linus construed it, into "write() returns -EBUSY" territory, which is just as bad but at least the writers aren't unkillable), and it reduces the I/O per splice significantly. You could limit each pipe to one outstanding splice and always leave Some space for normal I/O. This falls into "another lock just for this" territory I think, and it also sucks for the 99% of normal users.) I did this because Linus vaguely sanxioned it. It's probably feasible, but alas it isn't feasible for me. signature.asc Description: PGP signature
Re: [PATCH RERESEND 00/11] splice(file<>pipe) I/O on file as-if O_NONBLOCK
On 12/14/23 11:44 AM, Ahelenia Ziemia?ska wrote: > First: > https://lore.kernel.org/lkml/cover.1697486714.git.nabijaczlew...@nabijaczleweli.xyz/t/#u > Resend: > https://lore.kernel.org/lkml/1cover.1697486714.git.nabijaczlew...@nabijaczleweli.xyz/t/#u > Resending again per > https://lore.kernel.org/lkml/20231214093859.01f6e...@kernel.org/t/#u > > Hi! > > As it stands, splice(file -> pipe): > 1. locks the pipe, > 2. does a read from the file, > 3. unlocks the pipe. > > For reading from regular files and blcokdevs this makes no difference. > But if the file is a tty or a socket, for example, this means that until > data appears, which it may never do, every process trying to read from > or open the pipe enters an uninterruptible sleep, > and will only exit it if the splicing process is killed. > > This trivially denies service to: > * any hypothetical pipe-based log collexion system > * all nullmailer installations > * me, personally, when I'm pasting stuff into qemu -serial chardev:pipe > > This follows: > 1. > https://lore.kernel.org/linux-fsdevel/qk6hjuam54khlaikf2ssom6custxf5is2ekkaequf4hvode3ls@zgf7j5j4ubvw/t/#u > 2. a security@ thread rooted in > > 3. https://nabijaczleweli.xyz/content/blogn_t/011-linux-splice-exclusion.html > > Patches were posted and then discarded on principle or funxionality, > all in all terminating in Linus posting >> But it is possible that we need to just bite the bullet and say >> "copy_splice_read() needs to use a non-blocking kiocb for the IO". > > This does that, effectively making splice(file -> pipe) > request (and require) O_NONBLOCK on reads fron the file: > this doesn't affect splicing from regular files and blockdevs, > since they're always non-blocking > (and requesting the stronger "no kernel sleep" IOCB_NOWAIT is non-sensical), Not sure how you got the idea that regular files or block devices is always non-blocking, this is certainly not true without IOCB_NOWAIT. Without IOCB_NOWAIT, you can certainly be waiting for previous IO to complete. > but always returns -EINVAL for ttys. > Sockets behave as expected from O_NONBLOCK reads: > splice if there's data available else -EAGAIN. > > This should all pretty much behave as-expected. Should it? Seems like there's a very high risk of breaking existing use cases here. Have you at all looked into the approach of enabling splice to/from _without_ holding the pipe lock? That, to me, would seem like a much saner approach, with the caveat that I have not looked into that at all so there may indeed be reasons why this is not feasible. -- Jens Axboe
[PATCH RERESEND 00/11] splice(file<>pipe) I/O on file as-if O_NONBLOCK
First: https://lore.kernel.org/lkml/cover.1697486714.git.nabijaczlew...@nabijaczleweli.xyz/t/#u Resend: https://lore.kernel.org/lkml/1cover.1697486714.git.nabijaczlew...@nabijaczleweli.xyz/t/#u Resending again per https://lore.kernel.org/lkml/20231214093859.01f6e...@kernel.org/t/#u Hi! As it stands, splice(file -> pipe): 1. locks the pipe, 2. does a read from the file, 3. unlocks the pipe. For reading from regular files and blcokdevs this makes no difference. But if the file is a tty or a socket, for example, this means that until data appears, which it may never do, every process trying to read from or open the pipe enters an uninterruptible sleep, and will only exit it if the splicing process is killed. This trivially denies service to: * any hypothetical pipe-based log collexion system * all nullmailer installations * me, personally, when I'm pasting stuff into qemu -serial chardev:pipe This follows: 1. https://lore.kernel.org/linux-fsdevel/qk6hjuam54khlaikf2ssom6custxf5is2ekkaequf4hvode3ls@zgf7j5j4ubvw/t/#u 2. a security@ thread rooted in 3. https://nabijaczleweli.xyz/content/blogn_t/011-linux-splice-exclusion.html Patches were posted and then discarded on principle or funxionality, all in all terminating in Linus posting > But it is possible that we need to just bite the bullet and say > "copy_splice_read() needs to use a non-blocking kiocb for the IO". This does that, effectively making splice(file -> pipe) request (and require) O_NONBLOCK on reads fron the file: this doesn't affect splicing from regular files and blockdevs, since they're always non-blocking (and requesting the stronger "no kernel sleep" IOCB_NOWAIT is non-sensical), but always returns -EINVAL for ttys. Sockets behave as expected from O_NONBLOCK reads: splice if there's data available else -EAGAIN. This should all pretty much behave as-expected. Mostly a re-based version of the summary diff from . Bisexion yields commit 8924feff66f35fe22ce77aafe3f21eb8e5cff881 ("splice: lift pipe_lock out of splice_to_pipe()") as first bad. The patchset is made quite wide due to the many implementations of the splice_read callback, and was based entirely on results from $ git grep '\.splice_read.*=' | cut -d= -f2 | tr -s ',;[:space:]' '\n' | sort -u I'm assuming this is exhaustive, but it's 27 distinct implementations. Of these, I've classified these as trivial delegating wrappers: nfs_file_splice_read filemap_splice_read afs_file_splice_read filemap_splice_read ceph_splice_read filemap_splice_read ecryptfs_splice_read_update_atime filemap_splice_read ext4_file_splice_read filemap_splice_read f2fs_file_splice_read filemap_splice_read ntfs_file_splice_read filemap_splice_read ocfs2_file_splice_read filemap_splice_read orangefs_file_splice_read filemap_splice_read v9fs_file_splice_read filemap_splice_read xfs_file_splice_read filemap_splice_read zonefs_file_splice_readfilemap_splice_read sock_splice_read copy_splice_read or a socket-specific one coda_file_splice_read vfs_splice_read ovl_splice_readvfs_splice_read filemap_splice_read() is used for regular files and blockdevs, and thus needs no changes, and is thus unchanged. vfs_splice_read() delegates to copy_splice_read() or f_op->splice_read(). The rest are fixed, in patch order: 01. copy_splice_read() by simply doing the I/O with IOCB_NOWAIT; diff from Linus: https://lore.kernel.org/lkml/5osglsw36dla3mubtpsmdwdid4fsdacplyd6acx2igo4atogdg@yur3idyim3cc/t/#ee67de5a9ec18886c434113637d7eff6cd7acac4b 02. unix_stream_splice_read() by unconditionally passing MSG_DONTWAIT 03. fuse_dev_splice_read() by behaving as-if O_NONBLOCK 04. tracing_buffers_splice_read() by behaving as-if O_NONBLOCK (this also removes the retry loop) 05. relay_file_splice_read() by behaving as-if SPLICE_F_NONBLOCK (this just means EAGAINing unconditionally for an empty transfer) 06. smc_splice_read() by unconditionally passing MSG_DONTWAIT 07. kcm_splice_read() by unconditionally passing MSG_DONTWAIT 08. tls_sw_splice_read() by behaving as-if SPLICE_F_NONBLOCK 09. tcp_splice_read() by behaving as-if O_NONBLOCK (this also removes the retry loop) 10. EINVALs on files that neither have FMODE_NOWAIT nor are S_ISREG We don't want this to be just FMODE_NOWAIT since most regular files don't have it set and that's not the right semantic anyway, as noted at the top of this mail, But this allows blockdevs "by accident", effectively, since they have FMODE_NOWAIT (at least the ones I tried), even though they're just like regular files: handled by filemap_splice_read(), thus not dispatched with IOCB_NOWAIT. since always non-blocking. Should this be a check for FMODE_NOWAIT && (S_ISREG || S_ISBLK)? Should it remain FMODE_NOWAI