Re: [PATCH 5.7 233/244] RISC-V: Acquire mmap lock before invoking walk_page_range

[Stafford Horne]

On Wed, Jul 22, 2020 at 03:11:35PM +0000, Atish Patra wrote:
> On Wed, 2020-07-22 at 14:48 +0200, Greg KH wrote:
> > On Tue, Jul 21, 2020 at 03:50:35PM -0700, Palmer Dabbelt wrote:
> > > On Mon, 20 Jul 2020 12:14:03 PDT (-0700), Greg KH wrote:
> > > > On Mon, Jul 20, 2020 at 06:50:10PM +0000, Atish Patra wrote:
> > > > > On Mon, 2020-07-20 at 23:11 +0530, Naresh Kamboju wrote:
> > > > > > RISC-V build breaks on stable-rc 5.7 branch.
> > > > > > build failed with gcc-8, gcc-9 and gcc-9.
> > > > > > 
> > > > > 
> > > > > Sorry for the compilation issue.
> > > > > 
> > > > > mmap_read_lock was intrdouced in the following commit.
> > > > > 
> > > > > commit 9740ca4e95b4
> > > > > Author: Michel Lespinasse <walken@xxxxxxxxxx>
> > > > > Date:   Mon Jun 8 21:33:14 2020 -0700
> > > > > 
> > > > >     mmap locking API: initial implementation as rwsem wrappers
> > > > > 
> > > > > The following two commits replaced the usage of mmap_sem rwsem
> > > > > calls
> > > > > with mmap_lock.
> > > > > 
> > > > > d8ed45c5dcd4 (mmap locking API: use coccinelle to convert
> > > > > mmap_sem
> > > > > rwsem call sites)
> > > > > 89154dd5313f (mmap locking API: convert mmap_sem call sites
> > > > > missed by
> > > > > coccinelle)
> > > > > 
> > > > > The first commit is not present in stale 5.7-y for obvious
> > > > > reasons.
> > > > > 
> > > > > Do we need to send a separate patch only for stable branch with
> > > > > mmap_sem ? I am not sure if that will cause a conflict again in
> > > > > future.
> > > > 
> > > > I do not like taking odd backports, and would rather take the
> > > > real patch
> > > > that is upstream.
> > > 
> > > I guess I'm a bit new to stable backports so I'm not sure what's
> > > expected here.
> > > The failing patch fixes a bug by using a new interface.  The
> > > smallest diff fix
> > > for the stable kernels would be to construct a similar fix without
> > > the new
> > > interface, which in this case is very easy as the new interface
> > > just converted
> > > some generic locking calls to one-line functions.  It seems
> > > somewhat circuitous
> > > to land that in Linus' tree, though, as it would require breaking
> > > our port
> > > before fixing it to use the old interfaces and then cleaning it up
> > > to use the
> > > new interfaces.
> > > 
> > > Are we expected to pull the new interface onto stable in addition
> > > to this fix?
> > 
> > If it really does fix a big issue, yes, that is fine to do.
> > 
> 
> The original issue was manifests only for certain rootfs with
> CONFIG_DEBUG_VM enabled in kernel. I am not sure if that qualfies for a
> big issue :). But there was a similar fix for OpenRISC as well.
> 
> +stafford (who fixed the issue for OpenRISC)
> 
> @stafford Was there a request to backport the fix to stable ?

I have not requested  pulling my patch to stable.  Mine is this one:

  313a5257b84c2 ("openrisc: fix boot oops when DEBUG_VM is enabled")

If you cat request that would be great.

> I can combine all the git ids that needs to be pulled in.

Note, mine lists:

  Fixes: 42fc541404f2 ("mmap locking API: add mmap_assert_locked() and mmap_assert_write_locked()")

while your's lists:

  Fixes: 395a21ff859c(riscv: add ARCH_HAS_SET_DIRECT_MAP support)

That is when the code was introduced to the riscv port, but not the commit that
broke booting.

I think if you list the Fixes as I did when backporting to stable Greg, or his
tools, would also know that the patch depends on the the 42fc541404f2 commit.

Also, I guess there is no problem with listing 2 "Fixes" in the future.

Thanks Atish and Palmer.

-Stafford

> > > The new interface doesn't actually fix anything itself, but it
> > > would allow a
> > > functional kernel to be constructed that consisted of only
> > > backports from
> > > Linus' tree (which would also make further fixes easier).
> > 
> > That's fine.
> > 
> > > It seems safe to
> > > just pull in 9740ca4e95b4 ("mmap locking API: initial
> > > implementation as rwsem
> > > wrappers") before this failing patch, as in this case the new
> > > interface will
> > > function correctly with only a subset of callers having been
> > > converted.  Of
> > > course that's not a generally true statement so I don't know if
> > > future code
> > > will behave that way, but pulling in those conversion patches is
> > > definitely
> > > unnecessary diff right now.
> > 
> > If someone wants to send me a full set of the git ids that need to be
> > pulled in, I will be glad to do so.
> > 
> > thanks,
> > 
> > greg k-h
> 
> -- 
> Regards,
> Atish