Re: [PATCH] softmmu/physmem: Use qemu_madvise

[Andrew Deason]

On Tue, 22 Mar 2022 17:58:19 +0000
"Dr. David Alan Gilbert" <dgilbert@redhat.com> wrote:

> * Andrew Deason (adeason@sinenomine.net) wrote:
> > On Tue, 22 Mar 2022 16:53:05 +0000
> > "Dr. David Alan Gilbert" <dgilbert@redhat.com> wrote:
> > 
> > > * Andrew Deason (adeason@sinenomine.net) wrote:
> > > > On Wed, 16 Mar 2022 10:41:41 +0100
> > > > David Hildenbrand <david@redhat.com> wrote:
> > > > 
> > > > > On 16.03.22 10:37, Dr. David Alan Gilbert wrote:
> > > > > > * Peter Maydell (peter.maydell@linaro.org) wrote:
> > > > > >> On Wed, 16 Mar 2022 at 07:53, David Hildenbrand <david@redhat.com> 
> > > > > >> wrote:
> > > > > >>>
> > > > > >>> On 16.03.22 05:04, Andrew Deason wrote:
> > > > > >>>> We have a thin wrapper around madvise, called qemu_madvise, which
> > > > > >>>> provides consistent behavior for the !CONFIG_MADVISE case, and 
> > > > > >>>> works
> > > > > >>>> around some platform-specific quirks (some platforms only provide
> > > > > >>>> posix_madvise, and some don't offer all 'advise' types). This 
> > > > > >>>> specific
> > > > > >>>> caller of madvise has never used it, tracing back to its original
> > > > > >>>> introduction in commit e0b266f01dd2 ("migration_completion: Take
> > > > > >>>> current state").
> > > > > >>>>
> > > > > >>>> Call qemu_madvise here, to follow the same logic as all of our 
> > > > > >>>> other
> > > > > >>>> madvise callers. This slightly changes the behavior for
> > > > > >>>> !CONFIG_MADVISE (EINVAL instead of ENOSYS, and a slightly 
> > > > > >>>> different
> > > > > >>>> error message), but this is now more consistent with other 
> > > > > >>>> callers
> > > > > >>>> that use qemu_madvise.
> > > > > >>>>
> > > > > >>>> Signed-off-by: Andrew Deason <adeason@sinenomine.net>
> > > > > >>>> ---
> > > > > >>>> Looking at the history of commits that touch this madvise() 
> > > > > >>>> call, it
> > > > > >>>> doesn't _look_ like there's any reason to be directly calling 
> > > > > >>>> madvise vs
> > > > > >>>> qemu_advise (I don't see anything mentioned), but I'm not sure.
> > > > > >>>>
> > > > > >>>>  softmmu/physmem.c | 12 ++----------
> > > > > >>>>  1 file changed, 2 insertions(+), 10 deletions(-)
> > > > > >>>>
> > > > > >>>> diff --git a/softmmu/physmem.c b/softmmu/physmem.c
> > > > > >>>> index 43ae70fbe2..900c692b5e 100644
> > > > > >>>> --- a/softmmu/physmem.c
> > > > > >>>> +++ b/softmmu/physmem.c
> > > > > >>>> @@ -3584,40 +3584,32 @@ int ram_block_discard_range(RAMBlock 
> > > > > >>>> *rb, uint64_t start, size_t length)
> > > > > >>>>                           rb->idstr, start, length, ret);
> > > > > >>>>              goto err;
> > > > > >>>>  #endif
> > > > > >>>>          }
> > > > > >>>>          if (need_madvise) {
> > > > > >>>>              /* For normal RAM this causes it to be unmapped,
> > > > > >>>>               * for shared memory it causes the local mapping to 
> > > > > >>>> disappear
> > > > > >>>>               * and to fall back on the file contents (which we 
> > > > > >>>> just
> > > > > >>>>               * fallocate'd away).
> > > > > >>>>               */
> > > > > >>>> -#if defined(CONFIG_MADVISE)
> > > > > >>>>              if (qemu_ram_is_shared(rb) && rb->fd < 0) {
> > > > > >>>> -                ret = madvise(host_startaddr, length, 
> > > > > >>>> QEMU_MADV_REMOVE);
> > > > > >>>> +                ret = qemu_madvise(host_startaddr, length, 
> > > > > >>>> QEMU_MADV_REMOVE);
> > > > > >>>>              } else {
> > > > > >>>> -                ret = madvise(host_startaddr, length, 
> > > > > >>>> QEMU_MADV_DONTNEED);
> > > > > >>>> +                ret = qemu_madvise(host_startaddr, length, 
> > > > > >>>> QEMU_MADV_DONTNEED);
> > > > > >>>
> > > > > >>> posix_madvise(QEMU_MADV_DONTNEED) has completely different 
> > > > > >>> semantics
> > > > > >>> then madvise() -- it's not a discard that we need here.
> > > > > >>>
> > > > > >>> So ram_block_discard_range() would now succeed in environments 
> > > > > >>> (BSD?)
> > > > > >>> where it's supposed to fail.
> > > > > >>>
> > > > > >>> So AFAIKs this isn't sane.
> > > > > >>
> > > > > >> But CONFIG_MADVISE just means "host has madvise()"; it doesn't 
> > > > > >> imply
> > > > > >> "this is a Linux madvise() with MADV_DONTNEED". Solaris madvise()
> > > > > >> doesn't seem to have  MADV_DONTNEED at all; a quick look at the
> > > > > >> FreeBSD manpage suggests its madvise MADV_DONTNEED is identical
> > > > > >> to its posix_madvise MADV_DONTNEED.
> > > > > >>
> > > > > >> If we need "specifically Linux MADV_DONTNEED semantics" maybe we
> > > > > >> should define a QEMU_MADV_LINUX_DONTNEED which either (a) does the
> > > > > >> right thing or (b) fails, and use qemu_madvise() regardless.
> > > > > >>
> > > > > >> Certainly the current code is pretty fragile to being changed by
> > > > > >> people who don't understand the undocumented subtlety behind
> > > > > >> the use of a direct madvise() call here.
> > > > > > 
> > > > > > Yeh and I'm not sure I can remembe rall the subtleties; there's a 
> > > > > > big
> > > > > > hairy set of ifdef's in include/qemu/madvise.h that makes
> > > > > > sure we always have the definition of QEMU_MADV_REMOVE/DONTNEED
> > > > > > even on platforms that might not define it themselves.
> > > > > > 
> > > > > > But I think this code is used for things with different degrees
> > > > > > of care about the semantics; e.g. 'balloon' just cares that
> > > > > > it frees memory up and doesn't care about the detailed semantics
> > > > > > that much; so it's probably fine with that.
> > > > > > Postcopy is much more touchy, but then it's only going to be
> > > > > > calling this on Linux anyway (because of the userfault dependency).
> > > > > 
> > > > > MADV_DONTNEED/MADV_REMOVE only provides discard semantics on Linux 
> > > > > IIRC
> > > > > -- and that's what we want to achieve: ram_block_discard_range()
> > > > > 
> > > > > So I agree with Peter that we might want to make this more explicit.
> > > > 
> > > > I was looking at the comments/history around this code to try to make
> > > > this more explicit/clear, and it seems like the whole function is very
> > > > Linux-specific. All we ever do is:
> > > > 
> > > > - fallocate(FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE)
> > > > - madvise(MADV_REMOVE)
> > > > - madvise(MADV_DONTNEED) with Linux semantics
> > > > 
> > > > All of those operations are Linux-only, so trying to figure out the
> > > > cross-platform way to model this seems kind of pointless. Is it fine to
> > > > just #ifdef the whole thing to be just for Linux?
> > > 
> > > For ballooning we don't really need Linux semantics; we just need it to
> > > use less host memory.  Postcopy needs the more careful semantics though.
> > 
> > Right, sorry, you mentioned that, but I was thinking it sounded like
> > that applied to the MADV_DONTNEED path, which is only 1 case of the 3.
> > But reading into this a bit, maybe all of these cases are fine on
> > non-Linux for ballooning: fallocate is never called ('rb->fd' is always
> > -1), and QEMU_MADV_REMOVE falls back to QEMU_MADV_DONTNEED, which is
> > fine for ballooning. Am I understanding that correctly?
> > 
> > But regardless, it's simpler/more-conservative to make it all
> > Linux-specific. If I just take ram_block_discard_range away from
> > non-Linux (that is, make it always return an error), is that breaking
> > actual functionality, or is it removing a niche code path that non-Linux
> > isn't really using anyway, and it's not worth the time to figure out? I
> > am not familiar with qemu internals, so this is not obvious to me.
> > 
> > For context: I'm just trying to get the tree to compile on other
> > platforms (immediate focus is the guest agent on Solaris). The madvise()
> > calls here generate warnings due to platform-specific quirks that
> > qemu_madvise() has some logic to deal with. So my question is whether
> > to adapt these to the cross-platform qemu_advise(), or treat the
> > function as platform-specific code.
> 
> I think you should keep this function to have the basic functionality of
> freeing a chunk of memory so that it takes less host space up; that way
> a guest using virtio-balloon will still be nice to the host.

Okay, but the postcopy code path still requires the linux semantics
(that is, that accessing the range afterwards triggers a fault). That
sounds like having two functions is in order, to indicate which
semantics are needed? Something like:

- int ram_block_discard_range(...). Existing semantics; pages must be
  unmapped, or return an error. Called by postcopy stuff.

- void ram_block_try_discard_range(...). Best-effort, just try to reduce
  memory usage. Called by balloon stuff.

There are several callers in hw/virtio/virtio-mem.c that do look at the
return code. Do these require the pages be unmapped? (the linux
semantics)

-- 
Andrew Deason
adeason@sinenomine.net