From: Juergen Gross <jgross@suse.com>
To: qemu-devel@nongnu.org, xen-devel@lists.xenproject.org
Cc: anthony.perard@citrix.com, kraxel@redhat.com,
sstabellini@kernel.org, Juergen Gross <jgross@suse.com>
Subject: [Qemu-devel] [PATCH v2 2/2] xen: dont try setting max grants multiple times
Date: Fri, 22 Sep 2017 14:07:25 +0200 [thread overview]
Message-ID: <20170922120725.10910-3-jgross@suse.com> (raw)
In-Reply-To: <20170922120725.10910-1-jgross@suse.com>
Trying to call xengnttab_set_max_grants() with the same file handle
might fail on some kernels, as this operation is allowed only once.
This is a problem for the qdisk backend as blk_connect() can be
called multiple times for a domain, e.g. in case grub-xen is being
used to boot it.
So instead of letting the generic backend code open the gnttab device
do it in blk_connect() and close it again in blk_disconnect.
Signed-off-by: Juergen Gross <jgross@suse.com>
---
V2:
- always call blk_disconnect() from blk_free() in order to have the
gnttab device node closed (Anthony Perard)
---
hw/block/xen_disk.c | 16 ++++++++++++----
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/hw/block/xen_disk.c b/hw/block/xen_disk.c
index 62506e3167..e431bd89e8 100644
--- a/hw/block/xen_disk.c
+++ b/hw/block/xen_disk.c
@@ -1220,6 +1220,12 @@ static int blk_connect(struct XenDevice *xendev)
/* Add on the number needed for the ring pages */
max_grants += blkdev->nr_ring_ref;
+ blkdev->xendev.gnttabdev = xengnttab_open(NULL, 0);
+ if (blkdev->xendev.gnttabdev == NULL) {
+ xen_pv_printf(xendev, 0, "xengnttab_open failed: %s\n",
+ strerror(errno));
+ return -1;
+ }
if (xengnttab_set_max_grants(blkdev->xendev.gnttabdev, max_grants)) {
xen_pv_printf(xendev, 0, "xengnttab_set_max_grants failed: %s\n",
strerror(errno));
@@ -1327,6 +1333,11 @@ static void blk_disconnect(struct XenDevice *xendev)
}
blkdev->feature_persistent = false;
}
+
+ if (blkdev->xendev.gnttabdev) {
+ xengnttab_close(blkdev->xendev.gnttabdev);
+ blkdev->xendev.gnttabdev = NULL;
+ }
}
static int blk_free(struct XenDevice *xendev)
@@ -1334,9 +1345,7 @@ static int blk_free(struct XenDevice *xendev)
struct XenBlkDev *blkdev = container_of(xendev, struct XenBlkDev, xendev);
struct ioreq *ioreq;
- if (blkdev->blk || blkdev->sring) {
- blk_disconnect(xendev);
- }
+ blk_disconnect(xendev);
while (!QLIST_EMPTY(&blkdev->freelist)) {
ioreq = QLIST_FIRST(&blkdev->freelist);
@@ -1363,7 +1372,6 @@ static void blk_event(struct XenDevice *xendev)
struct XenDevOps xen_blkdev_ops = {
.size = sizeof(struct XenBlkDev),
- .flags = DEVOPS_FLAG_NEED_GNTDEV,
.alloc = blk_alloc,
.init = blk_init,
.initialise = blk_connect,
--
2.12.3
next prev parent reply other threads:[~2017-09-22 12:07 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-09-22 12:07 [Qemu-devel] [PATCH v2 0/2] xen: fix gnttab handling with old dom0 kernels Juergen Gross
2017-09-22 12:07 ` [Qemu-devel] [PATCH v2 1/2] xen: add a global indicator for grant copy being available Juergen Gross
2017-09-22 12:07 ` Juergen Gross [this message]
2017-09-25 14:41 ` [Qemu-devel] [PATCH v2 2/2] xen: dont try setting max grants multiple times Anthony PERARD
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=20170922120725.10910-3-jgross@suse.com \
--to=jgross@suse.com \
--cc=anthony.perard@citrix.com \
--cc=kraxel@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=sstabellini@kernel.org \
--cc=xen-devel@lists.xenproject.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).