[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] 9p: init_in_iov_from_pdu can truncate the size
From: |
Stefano Stabellini |
Subject: |
Re: [PATCH] 9p: init_in_iov_from_pdu can truncate the size |
Date: |
Thu, 19 Dec 2019 14:36:07 -0800 (PST) |
User-agent: |
Alpine 2.21 (DEB 202 2017-01-01) |
On Thu, 19 Dec 2019, Christian Schoenebeck wrote:
> On Donnerstag, 19. Dezember 2019 01:42:51 CET Stefano Stabellini wrote:
> > From: Stefano Stabellini <address@hidden>
> >
> > init_in_iov_from_pdu might not be able to allocate the full buffer size
> > requested, which comes from the client and could be larger than the
> > transport has available at the time of the request. Specifically, this
> > can happen with read operations, with the client requesting a read up to
> > the max allowed, which might be more than the transport has available at
> > the time.
>
> I haven't looked thoroughly at this yet, but that's about addressing a
> temporary, not a permanent transport buffer size limitation, right?
Yes, that is correct.
> Because if it was a permanent one, then probably an adjusted (lowered)
> msize should be returned on R_version response to client as well.
>
> I wonder why I never triggered this issue, because I was experimenting with
> huge msize values for 9pfs performance checks. Was there anything specific to
> trigger this issue?
Lots of heavy usage by a Java application booting. Nothing like Java to
stress the system :-)
> > Today the implementation of init_in_iov_from_pdu throws an error, both
> > Xen and Virtio.
> >
> > Instead, change the V9fsTransport interface so that the size becomes a
> > pointer and can be limited by the implementation of
> > init_in_iov_from_pdu.
> >
> > Change both the Xen and Virtio implementations to set the size to the
> > size of the buffer they managed to allocate, instead of throwing an
> > error.
> >
> > Signed-off-by: Stefano Stabellini <address@hidden>
> > CC: address@hidden
> > CC: address@hidden
> > ---
> > hw/9pfs/9p.c | 22 +++++++++++++++-------
> > hw/9pfs/9p.h | 2 +-
> > hw/9pfs/virtio-9p-device.c | 10 +++-------
> > hw/9pfs/xen-9p-backend.c | 12 ++++--------
> > 4 files changed, 23 insertions(+), 23 deletions(-)
> >
> > diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
> > index bdf7919abf..d6c89ce608 100644
> > --- a/hw/9pfs/9p.c
> > +++ b/hw/9pfs/9p.c
> > @@ -1682,22 +1682,30 @@ out_nofid:
> > * with qemu_iovec_destroy().
> > */
> > static void v9fs_init_qiov_from_pdu(QEMUIOVector *qiov, V9fsPDU *pdu,
> > - size_t skip, size_t size,
> > + size_t skip, size_t *size,
> > bool is_write)
> > {
> > QEMUIOVector elem;
> > struct iovec *iov;
> > unsigned int niov;
> > + size_t alloc_size = *size + skip;
> >
> > if (is_write) {
> > - pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov, size +
> > skip); + pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov,
> > alloc_size); } else {
> > - pdu->s->transport->init_in_iov_from_pdu(pdu, &iov, &niov, size +
> > skip); + pdu->s->transport->init_in_iov_from_pdu(pdu, &iov, &niov,
> > &alloc_size); + }
> > +
> > + if (alloc_size < skip)
> > + {
> > + *size = 0;
> > + } else {
> > + *size = alloc_size - skip;
> > }
> >
>
> Code style nitpicking:
>
> ERROR: that open brace { should be on the previous line
> #56: FILE: hw/9pfs/9p.c:1699:
> + if (alloc_size < skip)
> + {
Oops, sorry! I can fix that.
> >
> > static int v9fs_xattr_read(V9fsState *s, V9fsPDU *pdu, V9fsFidState *fidp,
> > @@ -1722,7 +1730,7 @@ static int v9fs_xattr_read(V9fsState *s, V9fsPDU *pdu,
> > V9fsFidState *fidp, }
> > offset += err;
> >
> > - v9fs_init_qiov_from_pdu(&qiov_full, pdu, offset, read_count, false);
> > + v9fs_init_qiov_from_pdu(&qiov_full, pdu, offset, &read_count, false);
> > err = v9fs_pack(qiov_full.iov, qiov_full.niov, 0,
> > ((char *)fidp->fs.xattr.value) + off,
> > read_count);
> > @@ -1852,7 +1860,7 @@ static void coroutine_fn v9fs_read(void *opaque)
> > QEMUIOVector qiov;
> > int32_t len;
> >
> > - v9fs_init_qiov_from_pdu(&qiov_full, pdu, offset + 4, max_count,
> > false); + v9fs_init_qiov_from_pdu(&qiov_full, pdu, offset + 4,
> > &max_count, false); qemu_iovec_init(&qiov, qiov_full.niov);
> > do {
> > qemu_iovec_reset(&qiov);
> > @@ -2085,7 +2093,7 @@ static void coroutine_fn v9fs_write(void *opaque)
> > return;
> > }
> > offset += err;
> > - v9fs_init_qiov_from_pdu(&qiov_full, pdu, offset, count, true);
> > + v9fs_init_qiov_from_pdu(&qiov_full, pdu, offset, &count, true);
> > trace_v9fs_write(pdu->tag, pdu->id, fid, off, count, qiov_full.niov);
> >
> > fidp = get_fid(pdu, fid);
> > diff --git a/hw/9pfs/9p.h b/hw/9pfs/9p.h
> > index 8883761b2c..50f7e21da6 100644
> > --- a/hw/9pfs/9p.h
> > +++ b/hw/9pfs/9p.h
> > @@ -365,7 +365,7 @@ struct V9fsTransport {
> > ssize_t (*pdu_vunmarshal)(V9fsPDU *pdu, size_t offset, const char
> > *fmt, va_list ap);
> > void (*init_in_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
> > - unsigned int *pniov, size_t size);
> > + unsigned int *pniov, size_t *size);
> > void (*init_out_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
> > unsigned int *pniov, size_t size); void (*push_and_notify)(V9fsPDU
> > *pdu);
> > diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
> > index 775e8ff766..68873c3f5f 100644
> > --- a/hw/9pfs/virtio-9p-device.c
> > +++ b/hw/9pfs/virtio-9p-device.c
> > @@ -145,19 +145,15 @@ static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu,
> > size_t offset, }
> >
> > static void virtio_init_in_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
> > - unsigned int *pniov, size_t size)
> > + unsigned int *pniov, size_t *size)
> > {
> > V9fsState *s = pdu->s;
> > V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
> > VirtQueueElement *elem = v->elems[pdu->idx];
> > size_t buf_size = iov_size(elem->in_sg, elem->in_num);
> >
> > - if (buf_size < size) {
> > - VirtIODevice *vdev = VIRTIO_DEVICE(v);
> > -
> > - virtio_error(vdev,
> > - "VirtFS reply type %d needs %zu bytes, buffer has
> > %zu", - pdu->id + 1, size, buf_size);
> > + if (buf_size < *size) {
> > + *size = buf_size;
> > }
> >
> > *piov = elem->in_sg;
> > diff --git a/hw/9pfs/xen-9p-backend.c b/hw/9pfs/xen-9p-backend.c
> > index 3f54a21c76..3994a356d4 100644
> > --- a/hw/9pfs/xen-9p-backend.c
> > +++ b/hw/9pfs/xen-9p-backend.c
> > @@ -187,7 +187,7 @@ static void xen_9pfs_init_out_iov_from_pdu(V9fsPDU *pdu,
> > static void xen_9pfs_init_in_iov_from_pdu(V9fsPDU *pdu,
> > struct iovec **piov,
> > unsigned int *pniov,
> > - size_t size)
> > + size_t *size)
> > {
> > Xen9pfsDev *xen_9pfs = container_of(pdu->s, Xen9pfsDev, state);
> > Xen9pfsRing *ring = &xen_9pfs->rings[pdu->tag % xen_9pfs->num_rings];
> > @@ -197,15 +197,11 @@ static void xen_9pfs_init_in_iov_from_pdu(V9fsPDU
> > *pdu, g_free(ring->sg);
> >
> > ring->sg = g_malloc0(sizeof(*ring->sg) * 2);
> > - xen_9pfs_in_sg(ring, ring->sg, &num, pdu->idx, size);
> > + xen_9pfs_in_sg(ring, ring->sg, &num, pdu->idx, *size);
> >
> > buf_size = iov_size(ring->sg, num);
> > - if (buf_size < size) {
> > - xen_pv_printf(&xen_9pfs->xendev, 0, "Xen 9pfs request type %d"
> > - "needs %zu bytes, buffer has %zu\n", pdu->id, size,
> > - buf_size);
> > - xen_be_set_state(&xen_9pfs->xendev, XenbusStateClosing);
> > - xen_9pfs_disconnect(&xen_9pfs->xendev);
> > + if (buf_size < *size) {
> > + *size = buf_size;
> > }
> >
> > *piov = ring->sg;
>
> Best regards,
> Christian Schoenebeck
>
>
>