From: Jan Kara <jack@suse.cz> To: rdodgen@gmail.com Cc: linux-ext4@vger.kernel.org, tytso@mit.edu, Randy Dodgen <dodgen@google.com>, linux-nvdimm@lists.01.org Subject: Re: [PATCH v3] Fix ext4 fault handling when mounted with -o dax,ro Date: Thu, 24 Aug 2017 16:45:44 +0200 [thread overview] Message-ID: <20170824144544.GE8969@quack2.suse.cz> (raw) In-Reply-To: <416a465a9fbe1d27085883dbf652c115cd195697.1503523424.git.dodgen@google.com> On Wed 23-08-17 14:26:52, rdodgen@gmail.com wrote: > From: Randy Dodgen <dodgen@google.com> > > If an ext4 filesystem is mounted with both the DAX and read-only > options, executables on that filesystem will fail to start (claiming > 'Segmentation fault') due to the fault handler returning > VM_FAULT_SIGBUS. > > This is due to the DAX fault handler (see ext4_dax_huge_fault) > attempting to write to the journal when FAULT_FLAG_WRITE is set. This is > the wrong behavior for write faults which will lead to a COW page; in > particular, this fails for readonly mounts. > > This change avoids journal writes for faults that are expected to COW. > > It might be the case that this could be better handled in > ext4_iomap_begin / ext4_iomap_end (called via iomap_ops inside > dax_iomap_fault). These is some overlap already (e.g. grabbing journal > handles). > > Signed-off-by: Randy Dodgen <dodgen@google.com> Thanks for the verbose comment :). The patch looks good to me. You can add: Reviewed-by: Jan Kara <jack@suse.cz> Honza > --- > > This version is simplified as suggested by Ross; all fault sizes and fallbacks > are handled by dax_iomap_fault. > > fs/ext4/file.c | 15 ++++++++++++++- > 1 file changed, 14 insertions(+), 1 deletion(-) > > diff --git a/fs/ext4/file.c b/fs/ext4/file.c > index 0d7cf0cc9b87..dc1e1fb6b54c 100644 > --- a/fs/ext4/file.c > +++ b/fs/ext4/file.c > @@ -279,7 +279,20 @@ static int ext4_dax_huge_fault(struct vm_fault *vmf, > handle_t *handle = NULL; > struct inode *inode = file_inode(vmf->vma->vm_file); > struct super_block *sb = inode->i_sb; > - bool write = vmf->flags & FAULT_FLAG_WRITE; > + > + /* > + * We have to distinguish real writes from writes which will result in a > + * COW page; COW writes should *not* poke the journal (the file will not > + * be changed). Doing so would cause unintended failures when mounted > + * read-only. > + * > + * We check for VM_SHARED rather than vmf->cow_page since the latter is > + * unset for pe_size != PE_SIZE_PTE (i.e. only in do_cow_fault); for > + * other sizes, dax_iomap_fault will handle splitting / fallback so that > + * we eventually come back with a COW page. > + */ > + bool write = (vmf->flags & FAULT_FLAG_WRITE) && > + (vmf->vma->vm_flags & VM_SHARED); > > if (write) { > sb_start_pagefault(sb); > -- > 2.14.1.342.g6490525c54-goog > -- Jan Kara <jack@suse.com> SUSE Labs, CR _______________________________________________ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
WARNING: multiple messages have this Message-ID (diff)
From: Jan Kara <jack-AlSwsSmVLrQ@public.gmane.org> To: rdodgen-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org Cc: linux-ext4-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, tytso-3s7WtUTddSA@public.gmane.org, Randy Dodgen <dodgen-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>, linux-nvdimm-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org Subject: Re: [PATCH v3] Fix ext4 fault handling when mounted with -o dax,ro Date: Thu, 24 Aug 2017 16:45:44 +0200 [thread overview] Message-ID: <20170824144544.GE8969@quack2.suse.cz> (raw) In-Reply-To: <416a465a9fbe1d27085883dbf652c115cd195697.1503523424.git.dodgen-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org> On Wed 23-08-17 14:26:52, rdodgen-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org wrote: > From: Randy Dodgen <dodgen-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org> > > If an ext4 filesystem is mounted with both the DAX and read-only > options, executables on that filesystem will fail to start (claiming > 'Segmentation fault') due to the fault handler returning > VM_FAULT_SIGBUS. > > This is due to the DAX fault handler (see ext4_dax_huge_fault) > attempting to write to the journal when FAULT_FLAG_WRITE is set. This is > the wrong behavior for write faults which will lead to a COW page; in > particular, this fails for readonly mounts. > > This change avoids journal writes for faults that are expected to COW. > > It might be the case that this could be better handled in > ext4_iomap_begin / ext4_iomap_end (called via iomap_ops inside > dax_iomap_fault). These is some overlap already (e.g. grabbing journal > handles). > > Signed-off-by: Randy Dodgen <dodgen-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org> Thanks for the verbose comment :). The patch looks good to me. You can add: Reviewed-by: Jan Kara <jack-AlSwsSmVLrQ@public.gmane.org> Honza > --- > > This version is simplified as suggested by Ross; all fault sizes and fallbacks > are handled by dax_iomap_fault. > > fs/ext4/file.c | 15 ++++++++++++++- > 1 file changed, 14 insertions(+), 1 deletion(-) > > diff --git a/fs/ext4/file.c b/fs/ext4/file.c > index 0d7cf0cc9b87..dc1e1fb6b54c 100644 > --- a/fs/ext4/file.c > +++ b/fs/ext4/file.c > @@ -279,7 +279,20 @@ static int ext4_dax_huge_fault(struct vm_fault *vmf, > handle_t *handle = NULL; > struct inode *inode = file_inode(vmf->vma->vm_file); > struct super_block *sb = inode->i_sb; > - bool write = vmf->flags & FAULT_FLAG_WRITE; > + > + /* > + * We have to distinguish real writes from writes which will result in a > + * COW page; COW writes should *not* poke the journal (the file will not > + * be changed). Doing so would cause unintended failures when mounted > + * read-only. > + * > + * We check for VM_SHARED rather than vmf->cow_page since the latter is > + * unset for pe_size != PE_SIZE_PTE (i.e. only in do_cow_fault); for > + * other sizes, dax_iomap_fault will handle splitting / fallback so that > + * we eventually come back with a COW page. > + */ > + bool write = (vmf->flags & FAULT_FLAG_WRITE) && > + (vmf->vma->vm_flags & VM_SHARED); > > if (write) { > sb_start_pagefault(sb); > -- > 2.14.1.342.g6490525c54-goog > -- Jan Kara <jack-IBi9RG/b67k@public.gmane.org> SUSE Labs, CR
next prev parent reply other threads:[~2017-08-24 14:43 UTC|newest] Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top 2017-08-22 22:24 [PATCH] Fix ext4 fault handling when mounted with -o dax,ro Randy Dodgen 2017-08-23 3:37 ` [PATCH v2] " rdodgen 2017-08-23 3:37 ` rdodgen 2017-08-23 16:38 ` Ross Zwisler 2017-08-23 16:38 ` Ross Zwisler 2017-08-23 20:11 ` Randy Dodgen 2017-08-23 20:11 ` Randy Dodgen 2017-08-23 21:26 ` [PATCH v3] " rdodgen 2017-08-23 21:26 ` rdodgen 2017-08-24 14:45 ` Jan Kara [this message] 2017-08-24 14:45 ` Jan Kara 2017-08-24 15:11 ` Ross Zwisler 2017-08-24 15:11 ` Ross Zwisler 2017-08-24 16:01 ` Christoph Hellwig 2017-08-24 16:01 ` Christoph Hellwig 2017-08-24 20:57 ` Theodore Ts'o 2017-08-24 20:57 ` Theodore Ts'o 2017-08-25 7:28 ` Christoph Hellwig 2017-08-25 7:28 ` Christoph Hellwig 2017-08-29 21:20 ` Christoph Hellwig 2017-08-29 21:20 ` Christoph Hellwig 2017-08-29 21:37 ` Ross Zwisler 2017-08-29 21:37 ` Ross Zwisler 2017-08-29 22:07 ` Randy Dodgen 2017-08-29 22:07 ` Randy Dodgen 2017-08-29 22:37 ` [fstests PATCH] generic: add test for executables on read-only DAX mounts Ross Zwisler 2017-08-29 22:37 ` Ross Zwisler 2017-08-30 10:59 ` Eryu Guan 2017-08-30 10:59 ` Eryu Guan 2017-08-31 4:01 ` Ross Zwisler 2017-08-31 4:01 ` Ross Zwisler 2017-08-31 4:01 ` Ross Zwisler 2017-08-31 4:09 ` [fstests v2] " Ross Zwisler 2017-08-31 4:09 ` Ross Zwisler 2017-08-30 14:51 ` [fstests PATCH] " Christoph Hellwig 2017-08-30 14:51 ` Christoph Hellwig 2017-08-30 14:51 ` Christoph Hellwig 2017-08-31 4:02 ` Ross Zwisler 2017-08-31 4:02 ` Ross Zwisler 2017-08-31 4:02 ` Ross Zwisler 2017-08-31 13:09 ` Christoph Hellwig 2017-08-31 13:09 ` Christoph Hellwig 2017-08-24 19:26 ` [PATCH v3] Fix ext4 fault handling when mounted with -o dax,ro Theodore Ts'o 2017-08-24 19:26 ` 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=20170824144544.GE8969@quack2.suse.cz \ --to=jack@suse.cz \ --cc=dodgen@google.com \ --cc=linux-ext4@vger.kernel.org \ --cc=linux-nvdimm@lists.01.org \ --cc=rdodgen@gmail.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.