All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Daniel P. Berrangé" <berrange@redhat.com>
To: David Hildenbrand <david@redhat.com>,
	Pankaj Gupta <pankaj.gupta.linux@gmail.com>,
	Eduardo Habkost <ehabkost@redhat.com>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	"Dr . David Alan Gilbert" <dgilbert@redhat.com>,
	qemu-devel@nongnu.org, Pankaj Gupta <pankaj.gupta@ionos.com>,
	Igor Mammedov <imammedo@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>
Subject: Re: [PATCH v3 6/7] util/oslib-posix: Support concurrent os_mem_prealloc() invocation
Date: Tue, 28 Sep 2021 18:01:03 +0100	[thread overview]
Message-ID: <YVNKT5sfSKD1nQXT@redhat.com> (raw)
In-Reply-To: <YVNJSYSP/IVUipc9@redhat.com>

On Tue, Sep 28, 2021 at 05:56:41PM +0100, Daniel P. Berrangé wrote:
> On Mon, Aug 16, 2021 at 11:47:38AM +0200, David Hildenbrand wrote:
> > Add a mutex to protect the SIGBUS case, as we cannot mess concurrently
> > with the sigbus handler and we have to manage the global variable
> > sigbus_memset_context. The MADV_POPULATE_WRITE path can run
> > concurrently.
> > 
> > Note that page_mutex and page_cond are shared between concurrent
> > invocations, which shouldn't be a problem.
> > 
> > This is a preparation for future virtio-mem prealloc code, which will call
> > os_mem_prealloc() asynchronously from an iothread when handling guest
> > requests.
> > 
> > Add a comment that messing with the SIGBUS handler is frowned upon and
> > can result in problems we fortunately haven't seen so far. Note that
> > forwarding signals to the already installed SIGBUS handler isn't clean
> > either, as that one might modify the SIGBUS handler again.
> 
> Even with the mutex, messing with SIGBUS post-startup still isn't safe
> as we're clashing with SIGBUS usage in softmmu/cpus.c
> 
> IIUC, the virtio-mem prealloc code is something new that we've not
> shipped yet. With this in mind, how about we simply enforce that
> usage of this new feature is dependant on kernel support for
> MADV_POPULATE_WRITE ?  If users want this feature they'll simply
> need to update to a modern kernel.  This shouldn't break any existing
> deployed QEMU guests IIUC

Oh, I should have read ahead to the next patch where you address
this issue.

With that in mind I'm ok adding

  Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>

> 
> > 
> > Reviewed-by: Pankaj Gupta <pankaj.gupta@ionos.com>
> > Signed-off-by: David Hildenbrand <david@redhat.com>
> > ---
> >  util/oslib-posix.c | 9 +++++++++
> >  1 file changed, 9 insertions(+)
> > 
> > diff --git a/util/oslib-posix.c b/util/oslib-posix.c
> > index efa4f96d56..9829149e4b 100644
> > --- a/util/oslib-posix.c
> > +++ b/util/oslib-posix.c
> > @@ -95,6 +95,7 @@ typedef struct MemsetThread MemsetThread;
> >  
> >  /* used by sigbus_handler() */
> >  static MemsetContext *sigbus_memset_context;
> > +static QemuMutex sigbus_mutex;
> >  
> >  static QemuMutex page_mutex;
> b>  static QemuCond page_cond;
> > @@ -625,6 +626,7 @@ static bool madv_populate_write_possible(char *area, size_t pagesize)
> >  void os_mem_prealloc(int fd, char *area, size_t memory, int smp_cpus,
> >                       Error **errp)
> >  {
> > +    static gsize initialized;
> >      int ret;
> >      struct sigaction act, oldact;
> >      size_t hpagesize = qemu_fd_getpagesize(fd);
> > @@ -638,6 +640,12 @@ void os_mem_prealloc(int fd, char *area, size_t memory, int smp_cpus,
> >      use_madv_populate_write = madv_populate_write_possible(area, hpagesize);
> >  
> >      if (!use_madv_populate_write) {
> > +        if (g_once_init_enter(&initialized)) {
> > +            qemu_mutex_init(&sigbus_mutex);
> > +            g_once_init_leave(&initialized, 1);
> > +        }
> > +
> > +        qemu_mutex_lock(&sigbus_mutex);
> >          memset(&act, 0, sizeof(act));
> >          act.sa_handler = &sigbus_handler;
> >          act.sa_flags = 0;
> > @@ -665,6 +673,7 @@ void os_mem_prealloc(int fd, char *area, size_t memory, int smp_cpus,
> >              perror("os_mem_prealloc: failed to reinstall signal handler");
> >              exit(1);
> >          }
> > +        qemu_mutex_unlock(&sigbus_mutex);
> >      }
> >  }
> >  
> > -- 
> > 2.31.1
> > 
> 
> Regards,
> Daniel
> -- 
> |: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
> |: https://libvirt.org         -o-            https://fstop138.berrange.com :|
> |: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
> 
> 

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|



  reply	other threads:[~2021-09-28 17:03 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-16  9:47 [PATCH v3 0/7] util/oslib-posix: Support MADV_POPULATE_WRITE for os_mem_prealloc() David Hildenbrand
2021-08-16  9:47 ` [PATCH v3 1/7] util/oslib-posix: Let touch_all_pages() return an error David Hildenbrand
2021-09-28 16:21   ` Daniel P. Berrangé
2021-08-16  9:47 ` [PATCH v3 2/7] util/oslib-posix: Support MADV_POPULATE_WRITE for os_mem_prealloc() David Hildenbrand
2021-09-28 16:45   ` Daniel P. Berrangé
2021-08-16  9:47 ` [PATCH v3 3/7] util/oslib-posix: Introduce and use MemsetContext for touch_all_pages() David Hildenbrand
2021-08-16  9:47 ` [PATCH v3 4/7] util/oslib-posix: Don't create too many threads with small memory or little pages David Hildenbrand
2021-09-28 16:47   ` Daniel P. Berrangé
2021-08-16  9:47 ` [PATCH v3 5/7] util/oslib-posix: Avoid creating a single thread with MADV_POPULATE_WRITE David Hildenbrand
2021-09-28 16:52   ` Daniel P. Berrangé
2021-08-16  9:47 ` [PATCH v3 6/7] util/oslib-posix: Support concurrent os_mem_prealloc() invocation David Hildenbrand
2021-09-28 16:56   ` Daniel P. Berrangé
2021-09-28 17:01     ` Daniel P. Berrangé [this message]
2021-09-28 17:17     ` David Hildenbrand
2021-08-16  9:47 ` [PATCH v3 7/7] util/oslib-posix: Forward SIGBUS to MCE handler under Linux David Hildenbrand
2021-09-28 16:59   ` Daniel P. Berrangé
2021-09-28 17:13     ` David Hildenbrand
2021-09-28 14:01 ` [PATCH v3 0/7] util/oslib-posix: Support MADV_POPULATE_WRITE for os_mem_prealloc() David Hildenbrand

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=YVNKT5sfSKD1nQXT@redhat.com \
    --to=berrange@redhat.com \
    --cc=david@redhat.com \
    --cc=dgilbert@redhat.com \
    --cc=ehabkost@redhat.com \
    --cc=imammedo@redhat.com \
    --cc=mst@redhat.com \
    --cc=pankaj.gupta.linux@gmail.com \
    --cc=pankaj.gupta@ionos.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /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: link
Be 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.