From: John Hubbard <jhubbard@nvidia.com> To: Theodore Ts'o <tytso@mit.edu> Cc: Eric Biggers <ebiggers@kernel.org>, Lee Jones <lee.jones@linaro.org>, linux-ext4@vger.kernel.org, Christoph Hellwig <hch@lst.de>, Dave Chinner <dchinner@redhat.com>, Goldwyn Rodrigues <rgoldwyn@suse.com>, "Darrick J . Wong" <darrick.wong@oracle.com>, Bob Peterson <rpeterso@redhat.com>, Damien Le Moal <damien.lemoal@wdc.com>, Andreas Gruenbacher <agruenba@redhat.com>, Ritesh Harjani <riteshh@linux.ibm.com>, Greg Kroah-Hartman <gregkh@linuxfoundation.org>, Johannes Thumshirn <jth@kernel.org>, linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org, cluster-devel@redhat.com, linux-kernel@vger.kernel.org Subject: Re: [PATCH -v3] ext4: don't BUG if kernel subsystems dirty pages without asking ext4 first Date: Fri, 25 Feb 2022 16:41:14 -0800 [thread overview] Message-ID: <303059e6-3a33-99cb-2952-82fe8079fa45@nvidia.com> (raw) In-Reply-To: <YhlkcYjozFmt3Kl4@mit.edu> On 2/25/22 15:21, Theodore Ts'o wrote: ... > For process_vm_writev() this is a case where user pages are pinned and > then released in short order, so I suspect that race with the page > cleaner would also be very hard to hit. But we could completely > remove the potential for the race, and also make things kinder for Completely removing the race would be wonderful. Because large supercomputer installations are good at hitting "rare" cases. > f2fs and btrfs's compressed file write support, by making things work > much like the write(2) system call. Imagine if we had a > "pin_user_pages_local()" which calls write_begin(), and a > "unpin_user_pages_local()" which calls write_end(), and the Right, that would supply the missing connection to the filesystems. In fact, maybe these names about right: pin_user_file_pages() unpin_user_file_pages() ...and then put them in a filesystem header file, because these are now tightly coupled to filesystems, what with the need to call .write_begin() and .write_end(). OK... > presumption with the "[un]pin_user_pages_local" API is that you don't > hold the pinned pages for very long --- say, not across a system call > boundary, and then it would work the same way the write(2) system call > works does except that in the case of process_vm_writev(2) the pages > are identified by another process's address space where they happen to > be mapped. > > This obviously doesn't work when pinning pages for remote DMA, because > in that case the time between pin_user_pages_remote() and > unpin_user_pages_remote() could be a long, long time, so that means we > can't use using write_begin/write_end; we'd need to call page_mkwrite() > when the pages are first pinned and then somehow prevent the page > cleaner from touching a dirty page which is pinned for use by the > remote DMA. > > Does that make sense? > > - Ted Yes, I really like this suggestion. It would neatly solve most short term pinning cases, without interfering with any future solutions for the long term pinning cases. Very nice. thanks, -- John Hubbard NVIDIA
WARNING: multiple messages have this Message-ID (diff)
From: John Hubbard <jhubbard@nvidia.com> To: cluster-devel.redhat.com Subject: [Cluster-devel] [PATCH -v3] ext4: don't BUG if kernel subsystems dirty pages without asking ext4 first Date: Fri, 25 Feb 2022 16:41:14 -0800 [thread overview] Message-ID: <303059e6-3a33-99cb-2952-82fe8079fa45@nvidia.com> (raw) In-Reply-To: <YhlkcYjozFmt3Kl4@mit.edu> On 2/25/22 15:21, Theodore Ts'o wrote: ... > For process_vm_writev() this is a case where user pages are pinned and > then released in short order, so I suspect that race with the page > cleaner would also be very hard to hit. But we could completely > remove the potential for the race, and also make things kinder for Completely removing the race would be wonderful. Because large supercomputer installations are good at hitting "rare" cases. > f2fs and btrfs's compressed file write support, by making things work > much like the write(2) system call. Imagine if we had a > "pin_user_pages_local()" which calls write_begin(), and a > "unpin_user_pages_local()" which calls write_end(), and the Right, that would supply the missing connection to the filesystems. In fact, maybe these names about right: pin_user_file_pages() unpin_user_file_pages() ...and then put them in a filesystem header file, because these are now tightly coupled to filesystems, what with the need to call .write_begin() and .write_end(). OK... > presumption with the "[un]pin_user_pages_local" API is that you don't > hold the pinned pages for very long --- say, not across a system call > boundary, and then it would work the same way the write(2) system call > works does except that in the case of process_vm_writev(2) the pages > are identified by another process's address space where they happen to > be mapped. > > This obviously doesn't work when pinning pages for remote DMA, because > in that case the time between pin_user_pages_remote() and > unpin_user_pages_remote() could be a long, long time, so that means we > can't use using write_begin/write_end; we'd need to call page_mkwrite() > when the pages are first pinned and then somehow prevent the page > cleaner from touching a dirty page which is pinned for use by the > remote DMA. > > Does that make sense? > > - Ted Yes, I really like this suggestion. It would neatly solve most short term pinning cases, without interfering with any future solutions for the long term pinning cases. Very nice. thanks, -- John Hubbard NVIDIA
next prev parent reply other threads:[~2022-02-26 0:41 UTC|newest] Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top 2022-02-16 16:31 [REPORT] kernel BUG at fs/ext4/inode.c:2620 - page_buffers() Lee Jones 2022-02-16 16:31 ` [Cluster-devel] " Lee Jones 2022-02-18 1:06 ` John Hubbard 2022-02-18 1:06 ` [Cluster-devel] " John Hubbard 2022-02-18 4:08 ` Theodore Ts'o 2022-02-18 4:08 ` [Cluster-devel] " Theodore Ts'o 2022-02-18 6:33 ` John Hubbard 2022-02-18 6:33 ` [Cluster-devel] " John Hubbard 2022-02-23 23:31 ` Theodore Ts'o 2022-02-23 23:31 ` [Cluster-devel] " Theodore Ts'o 2022-02-24 0:44 ` John Hubbard 2022-02-24 0:44 ` [Cluster-devel] " John Hubbard 2022-02-24 4:04 ` Theodore Ts'o 2022-02-24 4:04 ` [Cluster-devel] " Theodore Ts'o 2022-02-18 7:51 ` Greg Kroah-Hartman 2022-02-18 7:51 ` [Cluster-devel] " Greg Kroah-Hartman 2022-02-23 23:35 ` Theodore Ts'o 2022-02-23 23:35 ` [Cluster-devel] " Theodore Ts'o 2022-02-24 1:48 ` Dave Chinner 2022-02-24 1:48 ` [Cluster-devel] " Dave Chinner 2022-02-24 3:50 ` Theodore Ts'o 2022-02-24 3:50 ` [Cluster-devel] " Theodore Ts'o 2022-02-24 10:29 ` Dave Chinner 2022-02-24 10:29 ` [Cluster-devel] " Dave Chinner 2022-02-18 2:54 ` Theodore Ts'o 2022-02-18 2:54 ` [Cluster-devel] " Theodore Ts'o 2022-02-18 4:24 ` Matthew Wilcox 2022-02-18 4:24 ` [Cluster-devel] " Matthew Wilcox 2022-02-18 6:03 ` Theodore Ts'o 2022-02-18 6:03 ` [Cluster-devel] " Theodore Ts'o 2022-02-25 19:24 ` [PATCH -v2] ext4: don't BUG if kernel subsystems dirty pages without asking ext4 first Theodore Ts'o 2022-02-25 19:24 ` [Cluster-devel] " Theodore Ts'o 2022-02-25 20:51 ` Eric Biggers 2022-02-25 20:51 ` [Cluster-devel] " Eric Biggers 2022-02-25 21:08 ` Theodore Ts'o 2022-02-25 21:08 ` [Cluster-devel] " Theodore Ts'o 2022-02-25 21:23 ` [PATCH -v3] " Theodore Ts'o 2022-02-25 21:23 ` [Cluster-devel] " Theodore Ts'o 2022-02-25 21:33 ` John Hubbard 2022-02-25 21:33 ` [Cluster-devel] " John Hubbard 2022-02-25 23:21 ` Theodore Ts'o 2022-02-25 23:21 ` [Cluster-devel] " Theodore Ts'o 2022-02-26 0:18 ` Hillf Danton 2022-02-26 0:41 ` John Hubbard [this message] 2022-02-26 0:41 ` [Cluster-devel] " John Hubbard 2022-02-26 1:40 ` Theodore Ts'o 2022-02-26 1:40 ` [Cluster-devel] " Theodore Ts'o 2022-02-26 2:00 ` Theodore Ts'o 2022-02-26 2:00 ` [Cluster-devel] " Theodore Ts'o 2022-02-26 2:55 ` John Hubbard 2022-02-26 2:55 ` [Cluster-devel] " John Hubbard 2022-03-03 4:26 ` [PATCH -v4] " Theodore Ts'o 2022-03-03 4:26 ` [Cluster-devel] " Theodore Ts'o 2022-03-03 8:21 ` Christoph Hellwig 2022-03-03 8:21 ` [Cluster-devel] " Christoph Hellwig 2022-03-03 9:21 ` Lee Jones 2022-03-03 9:21 ` [Cluster-devel] " Lee Jones 2022-03-03 14:38 ` [PATCH -v5] ext4: don't BUG if someone " Theodore Ts'o 2022-03-03 14:38 ` [Cluster-devel] " Theodore Ts'o
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=303059e6-3a33-99cb-2952-82fe8079fa45@nvidia.com \ --to=jhubbard@nvidia.com \ --cc=agruenba@redhat.com \ --cc=cluster-devel@redhat.com \ --cc=damien.lemoal@wdc.com \ --cc=darrick.wong@oracle.com \ --cc=dchinner@redhat.com \ --cc=ebiggers@kernel.org \ --cc=gregkh@linuxfoundation.org \ --cc=hch@lst.de \ --cc=jth@kernel.org \ --cc=lee.jones@linaro.org \ --cc=linux-ext4@vger.kernel.org \ --cc=linux-fsdevel@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-xfs@vger.kernel.org \ --cc=rgoldwyn@suse.com \ --cc=riteshh@linux.ibm.com \ --cc=rpeterso@redhat.com \ --cc=tytso@mit.edu \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.