All of lore.kernel.org
 help / color / mirror / Atom feed
From: Samuel Thibault <samuel.thibault@eu.citrix.com>
To: xen-devel@lists.xensource.com
Subject: [PATCH] ioemu: improve DMA transfers
Date: Thu, 20 Mar 2008 16:29:15 +0000	[thread overview]
Message-ID: <20080320162915.GB11225@implementation.uk.xensource.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 2930 bytes --]

Hello,

I was wondering about the not so good performances of the IDE emulation
even in the case of DMA transfers.  It happens that qemu is actually
limiting the qemu / dom0 kernel reads/writes to 8KB at a time, which is
not very big (the SCSI emulation uses 64KB).  Increasing it provides a
very good improvement, as the attached graph shows.  On my test machine,
55MB/s can be achieved with a buffer of 128KB, to be compared to the
native 65MB/s.  The stubdomain approach is limited to 44KB because of
the limit in the PV blk interface.

Samuel


ioemu: improve DMA transfers
by increasing the size of DMA buffers.

Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>

diff -r dc1fe3bd1393 tools/ioemu/hw/ide.c
--- a/tools/ioemu/hw/ide.c	Thu Mar 20 10:53:10 2008 +0000
+++ b/tools/ioemu/hw/ide.c	Thu Mar 20 16:11:03 2008 +0000
@@ -189,6 +189,15 @@
 
 /* set to 1 set disable mult support */
 #define MAX_MULT_SECTORS 16
+#ifdef CONFIG_STUBDOM
+#include <xen/io/blkif.h>
+#define IDE_DMA_BUF_SIZE (BLKIF_MAX_SEGMENTS_PER_REQUEST * TARGET_PAGE_SIZE)
+#else
+#define IDE_DMA_BUF_SIZE 131072
+#endif
+#if (IDE_DMA_BUF_SIZE < MAX_MULT_SECTORS * 512)
+#error "IDE_DMA_BUF_SIZE must be bigger or equal to MAX_MULT_SECTORS * 512"
+#endif
 
 /* ATAPI defines */
 
@@ -932,8 +941,8 @@
 
     /* launch next transfer */
     n = s->nsector;
-    if (n > MAX_MULT_SECTORS)
-        n = MAX_MULT_SECTORS;
+    if (n > IDE_DMA_BUF_SIZE)
+        n = IDE_DMA_BUF_SIZE;
     s->io_buffer_index = 0;
     s->io_buffer_size = n * 512;
 #ifdef DEBUG_AIO
@@ -1041,8 +1050,8 @@
 
     /* launch next transfer */
     n = s->nsector;
-    if (n > MAX_MULT_SECTORS)
-        n = MAX_MULT_SECTORS;
+    if (n > IDE_DMA_BUF_SIZE)
+        n = IDE_DMA_BUF_SIZE;
     s->io_buffer_index = 0;
     s->io_buffer_size = n * 512;
 
@@ -1336,8 +1345,8 @@
         data_offset = 16;
     } else {
         n = s->packet_transfer_size >> 11;
-        if (n > (MAX_MULT_SECTORS / 4))
-            n = (MAX_MULT_SECTORS / 4);
+        if (n > (IDE_DMA_BUF_SIZE / 4))
+            n = (IDE_DMA_BUF_SIZE / 4);
         s->io_buffer_size = n * 2048;
         data_offset = 0;
     }
@@ -2306,7 +2315,7 @@
 
     for(i = 0; i < 2; i++) {
         s = ide_state + i;
-        s->io_buffer = qemu_memalign(getpagesize(), MAX_MULT_SECTORS*512 + 4);
+        s->io_buffer = qemu_memalign(getpagesize(), IDE_DMA_BUF_SIZE*512 + 4);
         if (i == 0)
             s->bs = hd0;
         else
--- a/tools/ioemu/hw/scsi-disk.c	Thu Mar 20 10:53:10 2008 +0000
+++ b/tools/ioemu/hw/scsi-disk.c	Thu Mar 20 16:11:03 2008 +0000
@@ -34,9 +34,10 @@
 #define SENSE_ILLEGAL_REQUEST 5
 
 #ifdef CONFIG_STUBDOM
-#define SCSI_DMA_BUF_SIZE    32768
+#include <xen/io/blkif.h>
+#define SCSI_DMA_BUF_SIZE    (BLKIF_MAX_SEGMENTS_PER_REQUEST * TARGET_PAGE_SIZE)
 #else
-#define SCSI_DMA_BUF_SIZE    65536
+#define SCSI_DMA_BUF_SIZE    131072
 #endif
 
 typedef struct SCSIRequest {

[-- Attachment #2: hda-perfs.eps --]
[-- Type: application/postscript, Size: 20188 bytes --]

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

             reply	other threads:[~2008-03-20 16:29 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-03-20 16:29 Samuel Thibault [this message]
2008-03-20 16:31 ` [PATCH] ioemu: improve DMA transfers Samuel Thibault
2008-03-20 16:39   ` Samuel Thibault

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=20080320162915.GB11225@implementation.uk.xensource.com \
    --to=samuel.thibault@eu.citrix.com \
    --cc=xen-devel@lists.xensource.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.