Message ID | 20201013103112.12132-1-xiubli@redhat.com |
---|---|
State | New |
Headers | show |
Series | ceph: add 'noshare' mount option support | expand |
On Tue, 2020-10-13 at 18:31 +0800, xiubli@redhat.com wrote: > From: Xiubo Li <xiubli@redhat.com> > > This will disable different mount points to share superblocks. > Why? What problem does this solve? Don't make us dig through random tracker bugs to determine this, please. :) Also, the subject mentions a "noshare" mount option, but the code below will be expecting sharesb/nosharesb. > URL: https://tracker.ceph.com/issues/46883 > Signed-off-by: Xiubo Li <xiubli@redhat.com> > --- > fs/ceph/super.c | 12 ++++++++++++ > fs/ceph/super.h | 1 + > 2 files changed, 13 insertions(+) > > diff --git a/fs/ceph/super.c b/fs/ceph/super.c > index 2f530a111b3a..6f283e4d62ee 100644 > --- a/fs/ceph/super.c > +++ b/fs/ceph/super.c > @@ -159,6 +159,7 @@ enum { > Opt_quotadf, > Opt_copyfrom, > Opt_wsync, > + Opt_sharesb, > }; > > enum ceph_recover_session_mode { > @@ -199,6 +200,7 @@ static const struct fs_parameter_spec ceph_mount_parameters[] = { > fsparam_string ("source", Opt_source), > fsparam_u32 ("wsize", Opt_wsize), > fsparam_flag_no ("wsync", Opt_wsync), > + fsparam_flag_no ("sharesb", Opt_sharesb), > {} > }; > > @@ -455,6 +457,12 @@ static int ceph_parse_mount_param(struct fs_context *fc, > else > fsopt->flags |= CEPH_MOUNT_OPT_ASYNC_DIROPS; > break; > + case Opt_sharesb: > + if (!result.negated) > + fsopt->flags &= ~CEPH_MOUNT_OPT_NO_SHARE_SB; > + else > + fsopt->flags |= CEPH_MOUNT_OPT_NO_SHARE_SB; > + break; > default: > BUG(); > } > @@ -1007,6 +1015,10 @@ static int ceph_compare_super(struct super_block *sb, struct fs_context *fc) > > dout("ceph_compare_super %p\n", sb); > > + if (fsopt->flags & CEPH_MOUNT_OPT_NO_SHARE_SB || > + other->mount_options->flags & CEPH_MOUNT_OPT_NO_SHARE_SB) > + return 0; > + > if (compare_mount_options(fsopt, opt, other)) { > dout("monitor(s)/mount options don't match\n"); > return 0; > diff --git a/fs/ceph/super.h b/fs/ceph/super.h > index f097237a5ad3..e877c21196e5 100644 > --- a/fs/ceph/super.h > +++ b/fs/ceph/super.h > @@ -44,6 +44,7 @@ > #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ > #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ > #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ > +#define CEPH_MOUNT_OPT_NO_SHARE_SB (1<<16) /* disable sharing the same superblock */ > > #define CEPH_MOUNT_OPT_DEFAULT \ > (CEPH_MOUNT_OPT_DCACHE | \ -- Jeff Layton <jlayton@kernel.org>
On 2020/10/13 20:31, Jeff Layton wrote: > On Tue, 2020-10-13 at 18:31 +0800, xiubli@redhat.com wrote: >> From: Xiubo Li <xiubli@redhat.com> >> >> This will disable different mount points to share superblocks. >> > Why? What problem does this solve? Don't make us dig through random > tracker bugs to determine this, please. :) So, should we just mannully trigger to crash the kernel to get the coredump ? > Also, the subject mentions a "noshare" mount option, but the code below > will be expecting sharesb/nosharesb. > >> URL: https://tracker.ceph.com/issues/46883 >> Signed-off-by: Xiubo Li <xiubli@redhat.com> >> --- >> fs/ceph/super.c | 12 ++++++++++++ >> fs/ceph/super.h | 1 + >> 2 files changed, 13 insertions(+) >> >> diff --git a/fs/ceph/super.c b/fs/ceph/super.c >> index 2f530a111b3a..6f283e4d62ee 100644 >> --- a/fs/ceph/super.c >> +++ b/fs/ceph/super.c >> @@ -159,6 +159,7 @@ enum { >> Opt_quotadf, >> Opt_copyfrom, >> Opt_wsync, >> + Opt_sharesb, >> }; >> >> enum ceph_recover_session_mode { >> @@ -199,6 +200,7 @@ static const struct fs_parameter_spec ceph_mount_parameters[] = { >> fsparam_string ("source", Opt_source), >> fsparam_u32 ("wsize", Opt_wsize), >> fsparam_flag_no ("wsync", Opt_wsync), >> + fsparam_flag_no ("sharesb", Opt_sharesb), >> {} >> }; >> >> @@ -455,6 +457,12 @@ static int ceph_parse_mount_param(struct fs_context *fc, >> else >> fsopt->flags |= CEPH_MOUNT_OPT_ASYNC_DIROPS; >> break; >> + case Opt_sharesb: >> + if (!result.negated) >> + fsopt->flags &= ~CEPH_MOUNT_OPT_NO_SHARE_SB; >> + else >> + fsopt->flags |= CEPH_MOUNT_OPT_NO_SHARE_SB; >> + break; >> default: >> BUG(); >> } >> @@ -1007,6 +1015,10 @@ static int ceph_compare_super(struct super_block *sb, struct fs_context *fc) >> >> dout("ceph_compare_super %p\n", sb); >> >> + if (fsopt->flags & CEPH_MOUNT_OPT_NO_SHARE_SB || >> + other->mount_options->flags & CEPH_MOUNT_OPT_NO_SHARE_SB) >> + return 0; >> + >> if (compare_mount_options(fsopt, opt, other)) { >> dout("monitor(s)/mount options don't match\n"); >> return 0; >> diff --git a/fs/ceph/super.h b/fs/ceph/super.h >> index f097237a5ad3..e877c21196e5 100644 >> --- a/fs/ceph/super.h >> +++ b/fs/ceph/super.h >> @@ -44,6 +44,7 @@ >> #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ >> #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ >> #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ >> +#define CEPH_MOUNT_OPT_NO_SHARE_SB (1<<16) /* disable sharing the same superblock */ >> >> #define CEPH_MOUNT_OPT_DEFAULT \ >> (CEPH_MOUNT_OPT_DCACHE | \
On Tue, 2020-10-13 at 20:44 +0800, Xiubo Li wrote: > On 2020/10/13 20:31, Jeff Layton wrote: > > On Tue, 2020-10-13 at 18:31 +0800, xiubli@redhat.com wrote: > > > From: Xiubo Li <xiubli@redhat.com> > > > > > > This will disable different mount points to share superblocks. > > > > > Why? What problem does this solve? Don't make us dig through random > > tracker bugs to determine this, please. :) > > So, should we just mannully trigger to crash the kernel to get the > coredump ? > Looking at the tracker ticket, it looks like you want this to work around some issues with lazy unmounts in tests. If so, then sure, forcing a coredump might give you an indication of what happened. In general though, this sounds like a hacky workaround for the problem in that tracker. I suggest we don't do this and work toward solving the real problems that caused the test writers to use lazy umounts in the first place... > > > Also, the subject mentions a "noshare" mount option, but the code below > > will be expecting sharesb/nosharesb. > > > > > URL: https://tracker.ceph.com/issues/46883 > > > Signed-off-by: Xiubo Li <xiubli@redhat.com> > > > --- > > > fs/ceph/super.c | 12 ++++++++++++ > > > fs/ceph/super.h | 1 + > > > 2 files changed, 13 insertions(+) > > > > > > diff --git a/fs/ceph/super.c b/fs/ceph/super.c > > > index 2f530a111b3a..6f283e4d62ee 100644 > > > --- a/fs/ceph/super.c > > > +++ b/fs/ceph/super.c > > > @@ -159,6 +159,7 @@ enum { > > > Opt_quotadf, > > > Opt_copyfrom, > > > Opt_wsync, > > > + Opt_sharesb, > > > }; > > > > > > enum ceph_recover_session_mode { > > > @@ -199,6 +200,7 @@ static const struct fs_parameter_spec ceph_mount_parameters[] = { > > > fsparam_string ("source", Opt_source), > > > fsparam_u32 ("wsize", Opt_wsize), > > > fsparam_flag_no ("wsync", Opt_wsync), > > > + fsparam_flag_no ("sharesb", Opt_sharesb), > > > {} > > > }; > > > > > > @@ -455,6 +457,12 @@ static int ceph_parse_mount_param(struct fs_context *fc, > > > else > > > fsopt->flags |= CEPH_MOUNT_OPT_ASYNC_DIROPS; > > > break; > > > + case Opt_sharesb: > > > + if (!result.negated) > > > + fsopt->flags &= ~CEPH_MOUNT_OPT_NO_SHARE_SB; > > > + else > > > + fsopt->flags |= CEPH_MOUNT_OPT_NO_SHARE_SB; > > > + break; > > > default: > > > BUG(); > > > } > > > @@ -1007,6 +1015,10 @@ static int ceph_compare_super(struct super_block *sb, struct fs_context *fc) > > > > > > dout("ceph_compare_super %p\n", sb); > > > > > > + if (fsopt->flags & CEPH_MOUNT_OPT_NO_SHARE_SB || > > > + other->mount_options->flags & CEPH_MOUNT_OPT_NO_SHARE_SB) > > > + return 0; > > > + > > > if (compare_mount_options(fsopt, opt, other)) { > > > dout("monitor(s)/mount options don't match\n"); > > > return 0; > > > diff --git a/fs/ceph/super.h b/fs/ceph/super.h > > > index f097237a5ad3..e877c21196e5 100644 > > > --- a/fs/ceph/super.h > > > +++ b/fs/ceph/super.h > > > @@ -44,6 +44,7 @@ > > > #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ > > > #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ > > > #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ > > > +#define CEPH_MOUNT_OPT_NO_SHARE_SB (1<<16) /* disable sharing the same superblock */ > > > > > > #define CEPH_MOUNT_OPT_DEFAULT \ > > > (CEPH_MOUNT_OPT_DCACHE | \ > >
On 2020/10/13 21:41, Jeff Layton wrote: > On Tue, 2020-10-13 at 20:44 +0800, Xiubo Li wrote: >> On 2020/10/13 20:31, Jeff Layton wrote: >>> On Tue, 2020-10-13 at 18:31 +0800, xiubli@redhat.com wrote: >>>> From: Xiubo Li <xiubli@redhat.com> >>>> >>>> This will disable different mount points to share superblocks. >>>> >>> Why? What problem does this solve? Don't make us dig through random >>> tracker bugs to determine this, please. :) >> So, should we just mannully trigger to crash the kernel to get the >> coredump ? >> > Looking at the tracker ticket, it looks like you want this to work > around some issues with lazy unmounts in tests. If so, then sure, > forcing a coredump might give you an indication of what happened. > > In general though, this sounds like a hacky workaround for the problem > in that tracker. I suggest we don't do this and work toward solving the > real problems that caused the test writers to use lazy umounts in the > first place... Okay, will try this. Thanks. >>> Also, the subject mentions a "noshare" mount option, but the code below >>> will be expecting sharesb/nosharesb. >>> >>>> URL: https://tracker.ceph.com/issues/46883 >>>> Signed-off-by: Xiubo Li <xiubli@redhat.com> >>>> --- >>>> fs/ceph/super.c | 12 ++++++++++++ >>>> fs/ceph/super.h | 1 + >>>> 2 files changed, 13 insertions(+) >>>> >>>> diff --git a/fs/ceph/super.c b/fs/ceph/super.c >>>> index 2f530a111b3a..6f283e4d62ee 100644 >>>> --- a/fs/ceph/super.c >>>> +++ b/fs/ceph/super.c >>>> @@ -159,6 +159,7 @@ enum { >>>> Opt_quotadf, >>>> Opt_copyfrom, >>>> Opt_wsync, >>>> + Opt_sharesb, >>>> }; >>>> >>>> enum ceph_recover_session_mode { >>>> @@ -199,6 +200,7 @@ static const struct fs_parameter_spec ceph_mount_parameters[] = { >>>> fsparam_string ("source", Opt_source), >>>> fsparam_u32 ("wsize", Opt_wsize), >>>> fsparam_flag_no ("wsync", Opt_wsync), >>>> + fsparam_flag_no ("sharesb", Opt_sharesb), >>>> {} >>>> }; >>>> >>>> @@ -455,6 +457,12 @@ static int ceph_parse_mount_param(struct fs_context *fc, >>>> else >>>> fsopt->flags |= CEPH_MOUNT_OPT_ASYNC_DIROPS; >>>> break; >>>> + case Opt_sharesb: >>>> + if (!result.negated) >>>> + fsopt->flags &= ~CEPH_MOUNT_OPT_NO_SHARE_SB; >>>> + else >>>> + fsopt->flags |= CEPH_MOUNT_OPT_NO_SHARE_SB; >>>> + break; >>>> default: >>>> BUG(); >>>> } >>>> @@ -1007,6 +1015,10 @@ static int ceph_compare_super(struct super_block *sb, struct fs_context *fc) >>>> >>>> dout("ceph_compare_super %p\n", sb); >>>> >>>> + if (fsopt->flags & CEPH_MOUNT_OPT_NO_SHARE_SB || >>>> + other->mount_options->flags & CEPH_MOUNT_OPT_NO_SHARE_SB) >>>> + return 0; >>>> + >>>> if (compare_mount_options(fsopt, opt, other)) { >>>> dout("monitor(s)/mount options don't match\n"); >>>> return 0; >>>> diff --git a/fs/ceph/super.h b/fs/ceph/super.h >>>> index f097237a5ad3..e877c21196e5 100644 >>>> --- a/fs/ceph/super.h >>>> +++ b/fs/ceph/super.h >>>> @@ -44,6 +44,7 @@ >>>> #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ >>>> #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ >>>> #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ >>>> +#define CEPH_MOUNT_OPT_NO_SHARE_SB (1<<16) /* disable sharing the same superblock */ >>>> >>>> #define CEPH_MOUNT_OPT_DEFAULT \ >>>> (CEPH_MOUNT_OPT_DCACHE | \ >>
diff --git a/fs/ceph/super.c b/fs/ceph/super.c index 2f530a111b3a..6f283e4d62ee 100644 --- a/fs/ceph/super.c +++ b/fs/ceph/super.c @@ -159,6 +159,7 @@ enum { Opt_quotadf, Opt_copyfrom, Opt_wsync, + Opt_sharesb, }; enum ceph_recover_session_mode { @@ -199,6 +200,7 @@ static const struct fs_parameter_spec ceph_mount_parameters[] = { fsparam_string ("source", Opt_source), fsparam_u32 ("wsize", Opt_wsize), fsparam_flag_no ("wsync", Opt_wsync), + fsparam_flag_no ("sharesb", Opt_sharesb), {} }; @@ -455,6 +457,12 @@ static int ceph_parse_mount_param(struct fs_context *fc, else fsopt->flags |= CEPH_MOUNT_OPT_ASYNC_DIROPS; break; + case Opt_sharesb: + if (!result.negated) + fsopt->flags &= ~CEPH_MOUNT_OPT_NO_SHARE_SB; + else + fsopt->flags |= CEPH_MOUNT_OPT_NO_SHARE_SB; + break; default: BUG(); } @@ -1007,6 +1015,10 @@ static int ceph_compare_super(struct super_block *sb, struct fs_context *fc) dout("ceph_compare_super %p\n", sb); + if (fsopt->flags & CEPH_MOUNT_OPT_NO_SHARE_SB || + other->mount_options->flags & CEPH_MOUNT_OPT_NO_SHARE_SB) + return 0; + if (compare_mount_options(fsopt, opt, other)) { dout("monitor(s)/mount options don't match\n"); return 0; diff --git a/fs/ceph/super.h b/fs/ceph/super.h index f097237a5ad3..e877c21196e5 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -44,6 +44,7 @@ #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ +#define CEPH_MOUNT_OPT_NO_SHARE_SB (1<<16) /* disable sharing the same superblock */ #define CEPH_MOUNT_OPT_DEFAULT \ (CEPH_MOUNT_OPT_DCACHE | \