From: "Michael S. Tsirkin" <mst@redhat.com>
To: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>,
linux-kernel@vger.kernel.org, Pawel Moll <pawel.moll@arm.com>,
virtio-dev@lists.oasis-open.org,
virtualization@lists.linux-foundation.org
Subject: Re: [PATCH v3 0/6] virtio_balloon: virtio 1 support
Date: Tue, 14 Apr 2015 11:58:32 +0200 [thread overview]
Message-ID: <20150414115440-mutt-send-email-mst@redhat.com> (raw)
In-Reply-To: <20150414115053.78189c71.cornelia.huck@de.ibm.com>
On Tue, Apr 14, 2015 at 11:50:53AM +0200, Cornelia Huck wrote:
> On Tue, 14 Apr 2015 11:21:11 +0200
> "Michael S. Tsirkin" <mst@redhat.com> wrote:
>
> > diff --git a/include/uapi/linux/virtio_balloon.h b/include/uapi/linux/virtio_balloon.h
> > index f81b220..164e0c2 100644
> > --- a/include/uapi/linux/virtio_balloon.h
> > +++ b/include/uapi/linux/virtio_balloon.h
> > @@ -52,15 +52,31 @@ struct virtio_balloon_config {
> > #define VIRTIO_BALLOON_S_MEMTOT 5 /* Total amount of memory */
> > #define VIRTIO_BALLOON_S_NR 6
> >
> > +/*
> > + * Memory statistics structure.
> > + * Driver fills an array of these structures and passes to device.
> > + *
> > + * NOTE: fields are laid out in a way that would make compiler add padding
> > + * between and after fields, so we have to use compiler-specific attributes to
> > + * pack it, to disable this padding. This also often causes compiler to
> > + * generate suboptimal code.
> > + *
> > + * We maintain this for backwards compatibility, but don't follow this example.
>
> s/this/the existing statistics structure/
existing seems redundant. What else? non-existing?
> > + *
> > + * Do something like the below instead:
>
> If you want to implement a similar structure, do...
>
> Just that nobody gets the idea that they are supposed to implement new
> balloon statistics ;)
>
> > + * struct virtio_balloon_stat {
> > + * __virtio16 tag;
> > + * __u8 reserved[6];
> > + * __virtio64 val;
> > + * };
>
> (...)
>
> > @@ -225,13 +219,8 @@ static inline void update_stat(struct virtio_balloon *vb, int idx,
> > u16 tag, u64 val)
> > {
> > BUG_ON(idx >= VIRTIO_BALLOON_S_NR);
> > - if (virtio_has_feature(vb->vdev, VIRTIO_F_VERSION_1)) {
> > - vb->stats[idx].tag = cpu_to_le32(tag);
> > - vb->stats[idx].val = cpu_to_le64(val);
> > - } else {
> > - vb->legacy_stats[idx].tag = tag;
> > - vb->legacy_stats[idx].val = val;
> > - }
> > + vb->stats[idx].tag = cpu_to_virtio16(vb->vdev, tag);
>
> Seems that nobody seemed to care much about statistics...
Or about BE guests ;)
> > + vb->stats[idx].val = cpu_to_virtio64(vb->vdev, val);
> > }
> >
> > #define pages_to_bytes(x) ((u64)(x) << PAGE_SHIFT)
> >
>
> With these changes merged in:
>
> Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
OK, here's an updated incremental patch: only comment
changed.
diff --git a/include/uapi/linux/virtio_balloon.h b/include/uapi/linux/virtio_balloon.h
index f81b220..984169a 100644
--- a/include/uapi/linux/virtio_balloon.h
+++ b/include/uapi/linux/virtio_balloon.h
@@ -52,15 +52,32 @@ struct virtio_balloon_config {
#define VIRTIO_BALLOON_S_MEMTOT 5 /* Total amount of memory */
#define VIRTIO_BALLOON_S_NR 6
+/*
+ * Memory statistics structure.
+ * Driver fills an array of these structures and passes to device.
+ *
+ * NOTE: fields are laid out in a way that would make compiler add padding
+ * between and after fields, so we have to use compiler-specific attributes to
+ * pack it, to disable this padding. This also often causes compiler to
+ * generate suboptimal code.
+ *
+ * We maintain this statistics structure format for backwards compatibility,
+ * but don't follow this example.
+ *
+ * If implementing a similar structure, do something like the below instead:
+ * struct virtio_balloon_stat {
+ * __virtio16 tag;
+ * __u8 reserved[6];
+ * __virtio64 val;
+ * };
+ *
+ * In other words, add explicit reserved fields to align field and
+ * structure boundaries at field size, avoiding compiler padding
+ * without the packed attribute.
+ */
struct virtio_balloon_stat {
- __u16 tag;
- __u64 val;
+ __virtio16 tag;
+ __virtio64 val;
} __attribute__((packed));
-struct virtio_balloon_stat_modern {
- __le16 tag;
- __u8 reserved[6];
- __le64 val;
-};
-
#endif /* _LINUX_VIRTIO_BALLOON_H */
diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloon.c
index 0583720..9db546e 100644
--- a/drivers/virtio/virtio_balloon.c
+++ b/drivers/virtio/virtio_balloon.c
@@ -77,10 +77,7 @@ struct virtio_balloon {
/* Memory statistics */
int need_stats_update;
- union {
- struct virtio_balloon_stat_modern stats[VIRTIO_BALLOON_S_NR];
- struct virtio_balloon_stat legacy_stats[VIRTIO_BALLOON_S_NR];
- };
+ struct virtio_balloon_stat stats[VIRTIO_BALLOON_S_NR];
/* To register callback in oom notifier call chain */
struct notifier_block nb;
@@ -93,10 +90,7 @@ static struct virtio_device_id id_table[] = {
static void stats_sg_init(struct virtio_balloon *vb, struct scatterlist *sg)
{
- if (virtio_has_feature(vb->vdev, VIRTIO_F_VERSION_1))
- sg_init_one(sg, vb->stats, sizeof(vb->stats));
- else
- sg_init_one(sg, vb->legacy_stats, sizeof(vb->legacy_stats));
+ sg_init_one(sg, vb->stats, sizeof(vb->stats));
}
static u32 page_to_balloon_pfn(struct page *page)
@@ -225,13 +219,8 @@ static inline void update_stat(struct virtio_balloon *vb, int idx,
u16 tag, u64 val)
{
BUG_ON(idx >= VIRTIO_BALLOON_S_NR);
- if (virtio_has_feature(vb->vdev, VIRTIO_F_VERSION_1)) {
- vb->stats[idx].tag = cpu_to_le32(tag);
- vb->stats[idx].val = cpu_to_le64(val);
- } else {
- vb->legacy_stats[idx].tag = tag;
- vb->legacy_stats[idx].val = val;
- }
+ vb->stats[idx].tag = cpu_to_virtio16(vb->vdev, tag);
+ vb->stats[idx].val = cpu_to_virtio64(vb->vdev, val);
}
#define pages_to_bytes(x) ((u64)(x) << PAGE_SHIFT)
next prev parent reply other threads:[~2015-04-14 9:58 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-04-01 12:57 [PATCH v3 0/6] virtio_balloon: virtio 1 support Michael S. Tsirkin
2015-04-01 10:35 ` [PATCH v3 1/6] virtio_balloon: transitional interface Michael S. Tsirkin
2015-04-01 10:49 ` Michael S. Tsirkin
2015-04-01 11:52 ` Cornelia Huck
2015-04-01 13:01 ` Michael S. Tsirkin
2015-04-01 10:35 ` [PATCH v3 2/6] virtio: balloon might not be a legacy device Michael S. Tsirkin
2015-04-01 11:57 ` Cornelia Huck
2015-04-01 10:35 ` [PATCH v3 3/6] virtio_ccw: support non-legacy balloon devices Michael S. Tsirkin
2015-04-01 10:40 ` [virtio-dev] " Christian Borntraeger
2015-04-01 11:55 ` Cornelia Huck
2015-04-14 1:09 ` Rusty Russell
2015-04-01 10:35 ` [PATCH v3 4/6] virtio_mmio: " Michael S. Tsirkin
2015-04-01 10:35 ` [PATCH v3 5/6] virtio_pci: " Michael S. Tsirkin
2015-04-01 10:36 ` [PATCH v3 6/6] virtio: drop virtio_device_is_legacy_only Michael S. Tsirkin
2015-04-01 12:04 ` Cornelia Huck
2015-04-12 15:02 ` [PATCH v3 0/6] virtio_balloon: virtio 1 support Michael S. Tsirkin
2015-04-14 1:12 ` Rusty Russell
2015-04-14 8:24 ` Cornelia Huck
2015-04-14 9:21 ` Michael S. Tsirkin
2015-04-14 9:50 ` Cornelia Huck
2015-04-14 9:58 ` Michael S. Tsirkin [this message]
2015-04-15 0:45 ` Rusty Russell
2015-04-15 15:32 ` Cornelia Huck
2015-04-15 15:45 ` Michael S. Tsirkin
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=20150414115440-mutt-send-email-mst@redhat.com \
--to=mst@redhat.com \
--cc=cornelia.huck@de.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=pawel.moll@arm.com \
--cc=rusty@rustcorp.com.au \
--cc=virtio-dev@lists.oasis-open.org \
--cc=virtualization@lists.linux-foundation.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).