Commit 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to
actual use case") made struct blkfront_ring_info size dynamic. This is
fine when running with only one queue, but with multiple queues the
addressing of the single queues has to be adapted as the structs are
allocated in an array.
Fixes: 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to actual use case")
Signed-off-by: Juergen Gross <[email protected]>
---
drivers/block/xen-blkfront.c | 82 ++++++++++++++++++++++++--------------------
1 file changed, 45 insertions(+), 37 deletions(-)
diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index e2ad6bba2281..a8d4a3838e5d 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -213,6 +213,7 @@ struct blkfront_info
struct blk_mq_tag_set tag_set;
struct blkfront_ring_info *rinfo;
unsigned int nr_rings;
+ unsigned int rinfo_size;
/* Save uncomplete reqs and bios for migration. */
struct list_head requests;
struct bio_list bio_list;
@@ -259,6 +260,21 @@ static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo);
static void blkfront_gather_backend_features(struct blkfront_info *info);
static int negotiate_mq(struct blkfront_info *info);
+#define rinfo_ptr(rinfo, off) \
+ (struct blkfront_ring_info *)((unsigned long)(rinfo) + (off))
+
+#define for_each_rinfo(info, rinfo, idx) \
+ for (rinfo = info->rinfo, idx = 0; \
+ idx < info->nr_rings; \
+ idx++, rinfo = rinfo_ptr(rinfo, info->rinfo_size))
+
+static struct blkfront_ring_info *get_rinfo(struct blkfront_info *info,
+ unsigned int i)
+{
+ BUG_ON(i >= info->nr_rings);
+ return rinfo_ptr(info->rinfo, i * info->rinfo_size);
+}
+
static int get_id_from_freelist(struct blkfront_ring_info *rinfo)
{
unsigned long free = rinfo->shadow_free;
@@ -883,8 +899,7 @@ static blk_status_t blkif_queue_rq(struct blk_mq_hw_ctx *hctx,
struct blkfront_info *info = hctx->queue->queuedata;
struct blkfront_ring_info *rinfo = NULL;
- BUG_ON(info->nr_rings <= qid);
- rinfo = &info->rinfo[qid];
+ rinfo = get_rinfo(info, qid);
blk_mq_start_request(qd->rq);
spin_lock_irqsave(&rinfo->ring_lock, flags);
if (RING_FULL(&rinfo->ring))
@@ -1181,6 +1196,7 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
static void xlvbd_release_gendisk(struct blkfront_info *info)
{
unsigned int minor, nr_minors, i;
+ struct blkfront_ring_info *rinfo;
if (info->rq == NULL)
return;
@@ -1188,8 +1204,7 @@ static void xlvbd_release_gendisk(struct blkfront_info *info)
/* No more blkif_request(). */
blk_mq_stop_hw_queues(info->rq);
- for (i = 0; i < info->nr_rings; i++) {
- struct blkfront_ring_info *rinfo = &info->rinfo[i];
+ for_each_rinfo(info, rinfo, i) {
/* No more gnttab callback work. */
gnttab_cancel_free_callback(&rinfo->callback);
@@ -1339,6 +1354,7 @@ static void blkif_free_ring(struct blkfront_ring_info *rinfo)
static void blkif_free(struct blkfront_info *info, int suspend)
{
unsigned int i;
+ struct blkfront_ring_info *rinfo;
/* Prevent new requests being issued until we fix things up. */
info->connected = suspend ?
@@ -1347,8 +1363,8 @@ static void blkif_free(struct blkfront_info *info, int suspend)
if (info->rq)
blk_mq_stop_hw_queues(info->rq);
- for (i = 0; i < info->nr_rings; i++)
- blkif_free_ring(&info->rinfo[i]);
+ for_each_rinfo(info, rinfo, i)
+ blkif_free_ring(rinfo);
kvfree(info->rinfo);
info->rinfo = NULL;
@@ -1775,6 +1791,7 @@ static int talk_to_blkback(struct xenbus_device *dev,
int err;
unsigned int i, max_page_order;
unsigned int ring_page_order;
+ struct blkfront_ring_info *rinfo;
if (!info)
return -ENODEV;
@@ -1788,9 +1805,7 @@ static int talk_to_blkback(struct xenbus_device *dev,
if (err)
goto destroy_blkring;
- for (i = 0; i < info->nr_rings; i++) {
- struct blkfront_ring_info *rinfo = &info->rinfo[i];
-
+ for_each_rinfo(info, rinfo, i) {
/* Create shared ring, alloc event channel. */
err = setup_blkring(dev, rinfo);
if (err)
@@ -1815,7 +1830,7 @@ static int talk_to_blkback(struct xenbus_device *dev,
/* We already got the number of queues/rings in _probe */
if (info->nr_rings == 1) {
- err = write_per_ring_nodes(xbt, &info->rinfo[0], dev->nodename);
+ err = write_per_ring_nodes(xbt, info->rinfo, dev->nodename);
if (err)
goto destroy_blkring;
} else {
@@ -1837,10 +1852,10 @@ static int talk_to_blkback(struct xenbus_device *dev,
goto abort_transaction;
}
- for (i = 0; i < info->nr_rings; i++) {
+ for_each_rinfo(info, rinfo, i) {
memset(path, 0, pathsize);
snprintf(path, pathsize, "%s/queue-%u", dev->nodename, i);
- err = write_per_ring_nodes(xbt, &info->rinfo[i], path);
+ err = write_per_ring_nodes(xbt, rinfo, path);
if (err) {
kfree(path);
goto destroy_blkring;
@@ -1868,9 +1883,8 @@ static int talk_to_blkback(struct xenbus_device *dev,
goto destroy_blkring;
}
- for (i = 0; i < info->nr_rings; i++) {
+ for_each_rinfo(info, rinfo, i) {
unsigned int j;
- struct blkfront_ring_info *rinfo = &info->rinfo[i];
for (j = 0; j < BLK_RING_SIZE(info); j++)
rinfo->shadow[j].req.u.rw.id = j + 1;
@@ -1900,6 +1914,7 @@ static int negotiate_mq(struct blkfront_info *info)
{
unsigned int backend_max_queues;
unsigned int i;
+ struct blkfront_ring_info *rinfo;
BUG_ON(info->nr_rings);
@@ -1911,20 +1926,16 @@ static int negotiate_mq(struct blkfront_info *info)
if (!info->nr_rings)
info->nr_rings = 1;
- info->rinfo = kvcalloc(info->nr_rings,
- struct_size(info->rinfo, shadow,
- BLK_RING_SIZE(info)),
- GFP_KERNEL);
+ info->rinfo_size = struct_size(info->rinfo, shadow,
+ BLK_RING_SIZE(info));
+ info->rinfo = kvcalloc(info->nr_rings, info->rinfo_size, GFP_KERNEL);
if (!info->rinfo) {
xenbus_dev_fatal(info->xbdev, -ENOMEM, "allocating ring_info structure");
info->nr_rings = 0;
return -ENOMEM;
}
- for (i = 0; i < info->nr_rings; i++) {
- struct blkfront_ring_info *rinfo;
-
- rinfo = &info->rinfo[i];
+ for_each_rinfo(info, rinfo, i) {
INIT_LIST_HEAD(&rinfo->indirect_pages);
INIT_LIST_HEAD(&rinfo->grants);
rinfo->dev_info = info;
@@ -2017,6 +2028,7 @@ static int blkif_recover(struct blkfront_info *info)
int rc;
struct bio *bio;
unsigned int segs;
+ struct blkfront_ring_info *rinfo;
blkfront_gather_backend_features(info);
/* Reset limits changed by blk_mq_update_nr_hw_queues(). */
@@ -2024,9 +2036,7 @@ static int blkif_recover(struct blkfront_info *info)
segs = info->max_indirect_segments ? : BLKIF_MAX_SEGMENTS_PER_REQUEST;
blk_queue_max_segments(info->rq, segs / GRANTS_PER_PSEG);
- for (r_index = 0; r_index < info->nr_rings; r_index++) {
- struct blkfront_ring_info *rinfo = &info->rinfo[r_index];
-
+ for_each_rinfo(info, rinfo, r_index) {
rc = blkfront_setup_indirect(rinfo);
if (rc)
return rc;
@@ -2036,10 +2046,7 @@ static int blkif_recover(struct blkfront_info *info)
/* Now safe for us to use the shared ring */
info->connected = BLKIF_STATE_CONNECTED;
- for (r_index = 0; r_index < info->nr_rings; r_index++) {
- struct blkfront_ring_info *rinfo;
-
- rinfo = &info->rinfo[r_index];
+ for_each_rinfo(info, rinfo, r_index) {
/* Kick any other new requests queued since we resumed */
kick_pending_request_queues(rinfo);
}
@@ -2072,13 +2079,13 @@ static int blkfront_resume(struct xenbus_device *dev)
struct blkfront_info *info = dev_get_drvdata(&dev->dev);
int err = 0;
unsigned int i, j;
+ struct blkfront_ring_info *rinfo;
dev_dbg(&dev->dev, "blkfront_resume: %s\n", dev->nodename);
bio_list_init(&info->bio_list);
INIT_LIST_HEAD(&info->requests);
- for (i = 0; i < info->nr_rings; i++) {
- struct blkfront_ring_info *rinfo = &info->rinfo[i];
+ for_each_rinfo(info, rinfo, i) {
struct bio_list merge_bio;
struct blk_shadow *shadow = rinfo->shadow;
@@ -2337,6 +2344,7 @@ static void blkfront_connect(struct blkfront_info *info)
unsigned int binfo;
char *envp[] = { "RESIZE=1", NULL };
int err, i;
+ struct blkfront_ring_info *rinfo;
switch (info->connected) {
case BLKIF_STATE_CONNECTED:
@@ -2394,8 +2402,8 @@ static void blkfront_connect(struct blkfront_info *info)
"physical-sector-size",
sector_size);
blkfront_gather_backend_features(info);
- for (i = 0; i < info->nr_rings; i++) {
- err = blkfront_setup_indirect(&info->rinfo[i]);
+ for_each_rinfo(info, rinfo, i) {
+ err = blkfront_setup_indirect(rinfo);
if (err) {
xenbus_dev_fatal(info->xbdev, err, "setup_indirect at %s",
info->xbdev->otherend);
@@ -2416,8 +2424,8 @@ static void blkfront_connect(struct blkfront_info *info)
/* Kick pending requests. */
info->connected = BLKIF_STATE_CONNECTED;
- for (i = 0; i < info->nr_rings; i++)
- kick_pending_request_queues(&info->rinfo[i]);
+ for_each_rinfo(info, rinfo, i)
+ kick_pending_request_queues(rinfo);
device_add_disk(&info->xbdev->dev, info->gd, NULL);
@@ -2652,9 +2660,9 @@ static void purge_persistent_grants(struct blkfront_info *info)
{
unsigned int i;
unsigned long flags;
+ struct blkfront_ring_info *rinfo;
- for (i = 0; i < info->nr_rings; i++) {
- struct blkfront_ring_info *rinfo = &info->rinfo[i];
+ for_each_rinfo(info, rinfo, i) {
struct grant *gnt_list_entry, *tmp;
spin_lock_irqsave(&rinfo->ring_lock, flags);
--
2.16.4
On Thu, Mar 05, 2020 at 11:03:31AM +0100, Juergen Gross wrote:
> Commit 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to
> actual use case") made struct blkfront_ring_info size dynamic. This is
> fine when running with only one queue, but with multiple queues the
> addressing of the single queues has to be adapted as the structs are
> allocated in an array.
Thanks, and sorry for not catching this during review.
>
> Fixes: 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to actual use case")
> Signed-off-by: Juergen Gross <[email protected]>
> ---
> drivers/block/xen-blkfront.c | 82 ++++++++++++++++++++++++--------------------
> 1 file changed, 45 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
> index e2ad6bba2281..a8d4a3838e5d 100644
> --- a/drivers/block/xen-blkfront.c
> +++ b/drivers/block/xen-blkfront.c
> @@ -213,6 +213,7 @@ struct blkfront_info
> struct blk_mq_tag_set tag_set;
> struct blkfront_ring_info *rinfo;
> unsigned int nr_rings;
> + unsigned int rinfo_size;
> /* Save uncomplete reqs and bios for migration. */
> struct list_head requests;
> struct bio_list bio_list;
> @@ -259,6 +260,21 @@ static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo);
> static void blkfront_gather_backend_features(struct blkfront_info *info);
> static int negotiate_mq(struct blkfront_info *info);
>
> +#define rinfo_ptr(rinfo, off) \
> + (struct blkfront_ring_info *)((unsigned long)(rinfo) + (off))
^ void * would seem more natural IMO.
Also if you use void * you don't need the extra (struct
blkfront_ring_info *) cast I think?
I however think this macro is kind of weird, since it's just doing an
addition. I would rather have that calculation in get_rinfo and code
for_each_rinfo on top of that.
I agree this might be a question of taste, so I'm not going to insist
but that would reduce the number of helpers from 3 to 2.
> +
> +#define for_each_rinfo(info, rinfo, idx) \
> + for (rinfo = info->rinfo, idx = 0; \
> + idx < info->nr_rings; \
> + idx++, rinfo = rinfo_ptr(rinfo, info->rinfo_size))
I think the above is missing proper parentheses around macro
parameters.
> +
> +static struct blkfront_ring_info *get_rinfo(struct blkfront_info *info,
> + unsigned int i)
inline attribute might be appropriate here.
Thanks, Roger.
On 05.03.20 11:49, Roger Pau Monné wrote:
> On Thu, Mar 05, 2020 at 11:03:31AM +0100, Juergen Gross wrote:
>> Commit 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to
>> actual use case") made struct blkfront_ring_info size dynamic. This is
>> fine when running with only one queue, but with multiple queues the
>> addressing of the single queues has to be adapted as the structs are
>> allocated in an array.
>
> Thanks, and sorry for not catching this during review.
>
>>
>> Fixes: 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to actual use case")
>> Signed-off-by: Juergen Gross <[email protected]>
>> ---
>> drivers/block/xen-blkfront.c | 82 ++++++++++++++++++++++++--------------------
>> 1 file changed, 45 insertions(+), 37 deletions(-)
>>
>> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
>> index e2ad6bba2281..a8d4a3838e5d 100644
>> --- a/drivers/block/xen-blkfront.c
>> +++ b/drivers/block/xen-blkfront.c
>> @@ -213,6 +213,7 @@ struct blkfront_info
>> struct blk_mq_tag_set tag_set;
>> struct blkfront_ring_info *rinfo;
>> unsigned int nr_rings;
>> + unsigned int rinfo_size;
>> /* Save uncomplete reqs and bios for migration. */
>> struct list_head requests;
>> struct bio_list bio_list;
>> @@ -259,6 +260,21 @@ static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo);
>> static void blkfront_gather_backend_features(struct blkfront_info *info);
>> static int negotiate_mq(struct blkfront_info *info);
>>
>> +#define rinfo_ptr(rinfo, off) \
>> + (struct blkfront_ring_info *)((unsigned long)(rinfo) + (off))
> ^ void * would seem more natural IMO.
>
> Also if you use void * you don't need the extra (struct
> blkfront_ring_info *) cast I think?
Yes, can change that.
> I however think this macro is kind of weird, since it's just doing an
> addition. I would rather have that calculation in get_rinfo and code
> for_each_rinfo on top of that.
I wanted to avoid the multiplication in the rather common
for_each_rinfo() usage.
>
> I agree this might be a question of taste, so I'm not going to insist
> but that would reduce the number of helpers from 3 to 2.
>
>> +
>> +#define for_each_rinfo(info, rinfo, idx) \
>> + for (rinfo = info->rinfo, idx = 0; \
>> + idx < info->nr_rings; \
>> + idx++, rinfo = rinfo_ptr(rinfo, info->rinfo_size))
>
> I think the above is missing proper parentheses around macro
> parameters.
rinfo and idx are simple variables, so I don't think they need
parentheses. info maybe. But just seeing it now: naming the
parameter "rinfo" and trying to access info->rinfo isn't a good
idea. It is working only as I always use "rinfo" as the pointer.
>
>> +
>> +static struct blkfront_ring_info *get_rinfo(struct blkfront_info *info,
>> + unsigned int i)
>
> inline attribute might be appropriate here.
See "the inline disease" in the kernel's coding style.
Juergen
On Thu, Mar 05, 2020 at 12:04:27PM +0100, Jürgen Groß wrote:
> On 05.03.20 11:49, Roger Pau Monné wrote:
> > On Thu, Mar 05, 2020 at 11:03:31AM +0100, Juergen Gross wrote:
> > > Commit 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to
> > > actual use case") made struct blkfront_ring_info size dynamic. This is
> > > fine when running with only one queue, but with multiple queues the
> > > addressing of the single queues has to be adapted as the structs are
> > > allocated in an array.
> >
> > Thanks, and sorry for not catching this during review.
> >
> > >
> > > Fixes: 0265d6e8ddb890 ("xen/blkfront: limit allocated memory size to actual use case")
> > > Signed-off-by: Juergen Gross <[email protected]>
> > > ---
> > > drivers/block/xen-blkfront.c | 82 ++++++++++++++++++++++++--------------------
> > > 1 file changed, 45 insertions(+), 37 deletions(-)
> > >
> > > diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
> > > index e2ad6bba2281..a8d4a3838e5d 100644
> > > --- a/drivers/block/xen-blkfront.c
> > > +++ b/drivers/block/xen-blkfront.c
> > > @@ -213,6 +213,7 @@ struct blkfront_info
> > > struct blk_mq_tag_set tag_set;
> > > struct blkfront_ring_info *rinfo;
> > > unsigned int nr_rings;
> > > + unsigned int rinfo_size;
> > > /* Save uncomplete reqs and bios for migration. */
> > > struct list_head requests;
> > > struct bio_list bio_list;
> > > @@ -259,6 +260,21 @@ static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo);
> > > static void blkfront_gather_backend_features(struct blkfront_info *info);
> > > static int negotiate_mq(struct blkfront_info *info);
> > > +#define rinfo_ptr(rinfo, off) \
> > > + (struct blkfront_ring_info *)((unsigned long)(rinfo) + (off))
> > ^ void * would seem more natural IMO.
> >
> > Also if you use void * you don't need the extra (struct
> > blkfront_ring_info *) cast I think?
>
> Yes, can change that.
>
> > I however think this macro is kind of weird, since it's just doing an
> > addition. I would rather have that calculation in get_rinfo and code
> > for_each_rinfo on top of that.
>
> I wanted to avoid the multiplication in the rather common
> for_each_rinfo() usage.
Can you undef it afterwards then? I don't think it's supposed to be
used by the rest of the file.
>
> >
> > I agree this might be a question of taste, so I'm not going to insist
> > but that would reduce the number of helpers from 3 to 2.
> >
> > > +
> > > +#define for_each_rinfo(info, rinfo, idx) \
> > > + for (rinfo = info->rinfo, idx = 0; \
> > > + idx < info->nr_rings; \
> > > + idx++, rinfo = rinfo_ptr(rinfo, info->rinfo_size))
> >
> > I think the above is missing proper parentheses around macro
> > parameters.
>
> rinfo and idx are simple variables, so I don't think they need
> parentheses. info maybe. But just seeing it now: naming the
> parameter "rinfo" and trying to access info->rinfo isn't a good
> idea. It is working only as I always use "rinfo" as the pointer.
Dereferences of info and the increase of idx should have parentheses
IMO.
You could rename the rinfo parameter to entry or some such.
> >
> > > +
> > > +static struct blkfront_ring_info *get_rinfo(struct blkfront_info *info,
> > > + unsigned int i)
> >
> > inline attribute might be appropriate here.
>
> See "the inline disease" in the kernel's coding style.
This function has two lines, so I think it's suitable to be inlined:
"A reasonable rule of thumb is to not put inline at functions that
have more than 3 lines of code in them"
I bet the compiler would do this already, but I think adding inline
here is fine according to coding style.
Thanks, Roger.