Skip to contentSkip to conversation

xen/blkfront: increase the default value of xen_blkif_max_segments

PATCH RESEND In review Bob Liu 2 participants 3 messages Jan 21, 2015

Subsystems (3)
lore
Share & tools
lore

Bob Liu

author
#0 lore raw

PATCH RESEND 1/1 xen/blkfront: increase the default value of xen_blkif_max_segments

The default max-segments of indirect requests was 32, so IO operations with bigger block size(>32*4k) would be split and make performance drop.

Nowadays backend device usually support 512k max_sectors_kb on desktop, and usually larger on server machines connected with high-end storage system. The default max size(128k) is not appropriate anymore, this patch increases it to 128(128*4k=512k).

---
 drivers/block/xen-blkfront.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/block/xen-blkfront.c +2 -2
index 2236c6f..1bf2429 100644
@@ -94,9 +94,9 @@ static const struct block_device_operations xlvbd_block_fops;
  * by the backend driver.
  */
 
-static unsigned int xen_blkif_max_segments = 32;
+static unsigned int xen_blkif_max_segments = 128;
 module_param_named(max, xen_blkif_max_segments, int, S_IRUGO);
-MODULE_PARM_DESC(max, "Maximum amount of segments in indirect requests (default is 32)");
+MODULE_PARM_DESC(max, "Maximum amount of segments in indirect requests (default is 128)");
 
 #define BLK_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE)
 
Signature · 1 line
1.7.10.4
#1 lore raw

El 21/01/15 a les 8.56, Bob Liu ha escrit:

Quoted 9 lines

The default max-segments of indirect requests was 32, so IO operations with bigger block size(>32*4k) would be split and make performance drop.

Nowadays backend device usually support 512k max_sectors_kb on desktop, and usually larger on server machines connected with high-end storage system. The default max size(128k) is not appropriate anymore, this patch increases it to 128(128*4k=512k).

Looks fine to me, the Linux backend is also prepared to handle it without any modifications. As I said in the last version, I would add the following to the commit message:

This change implies we can now have 32*128+32 = 4128 in-flight grants, which greatly surpasses the default amount of persistent grants blkback can handle, so the LRU in blkback will kick in.

Roger.

#2 lore raw
In reply to Bob Liu

El 21/01/15 a les 8.56, Bob Liu ha escrit:

Quoted 9 lines

The default max-segments of indirect requests was 32, so IO operations with bigger block size(>32*4k) would be split and make performance drop.

Nowadays backend device usually support 512k max_sectors_kb on desktop, and usually larger on server machines connected with high-end storage system. The default max size(128k) is not appropriate anymore, this patch increases it to 128(128*4k=512k).

Looks fine to me, the Linux backend is also prepared to handle it without any modifications. As I said in the last version, I would add the following to the commit message:

This change implies we can now have 32*128+32 = 4128 in-flight grants, which greatly surpasses the default amount of persistent grants blkback can handle, so the LRU in blkback will kick in.

Roger.