[PATCH] IB/ipath: use more appropriate gfp flags
authorBryan O'Sullivan <bos@pathscale.com>
Sat, 1 Jul 2006 11:36:01 +0000 (04:36 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Sat, 1 Jul 2006 16:56:00 +0000 (09:56 -0700)
This helps us to survive better when memory is fragmented.

Signed-off-by: Dave Olson <dave.olson@qlogic.com>
Signed-off-by: Bryan O'Sullivan <bryan.osullivan@qlogic.com>
Cc: "Michael S. Tsirkin" <mst@mellanox.co.il>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
drivers/infiniband/hw/ipath/ipath_file_ops.c

index cd2d5cd..03689db 100644 (file)
@@ -705,6 +705,15 @@ static int ipath_create_user_egr(struct ipath_portdata *pd)
        unsigned e, egrcnt, alloced, egrperchunk, chunk, egrsize, egroff;
        size_t size;
        int ret;
+       gfp_t gfp_flags;
+
+       /*
+        * GFP_USER, but without GFP_FS, so buffer cache can be
+        * coalesced (we hope); otherwise, even at order 4,
+        * heavy filesystem activity makes these fail, and we can
+        * use compound pages.
+        */
+       gfp_flags = __GFP_WAIT | __GFP_IO | __GFP_COMP;
 
        egrcnt = dd->ipath_rcvegrcnt;
        /* TID number offset for this port */
@@ -721,10 +730,8 @@ static int ipath_create_user_egr(struct ipath_portdata *pd)
         * memory pressure (creating large files and then copying them over
         * NFS while doing lots of MPI jobs), we hit some allocation
         * failures, even though we can sleep...  (2.6.10) Still get
-        * failures at 64K.  32K is the lowest we can go without waiting
-        * more memory again.  It seems likely that the coalescing in
-        * free_pages, etc. still has issues (as it has had previously
-        * during 2.6.x development).
+        * failures at 64K.  32K is the lowest we can go without wasting
+        * additional memory.
         */
        size = 0x8000;
        alloced = ALIGN(egrsize * egrcnt, size);
@@ -745,12 +752,6 @@ static int ipath_create_user_egr(struct ipath_portdata *pd)
                goto bail_rcvegrbuf;
        }
        for (e = 0; e < pd->port_rcvegrbuf_chunks; e++) {
-               /*
-                * GFP_USER, but without GFP_FS, so buffer cache can be
-                * coalesced (we hope); otherwise, even at order 4,
-                * heavy filesystem activity makes these fail
-                */
-               gfp_t gfp_flags = __GFP_WAIT | __GFP_IO | __GFP_COMP;
 
                pd->port_rcvegrbuf[e] = dma_alloc_coherent(
                        &dd->pcidev->dev, size, &pd->port_rcvegrbuf_phys[e],
@@ -1167,9 +1168,10 @@ static int ipath_mmap(struct file *fp, struct vm_area_struct *vma)
 
        ureg = dd->ipath_uregbase + dd->ipath_palign * pd->port_port;
 
-       ipath_cdbg(MM, "ushare: pgaddr %llx vm_start=%lx, vmlen %lx\n",
+       ipath_cdbg(MM, "pgaddr %llx vm_start=%lx len %lx port %u:%u\n",
                   (unsigned long long) pgaddr, vma->vm_start,
-                  vma->vm_end - vma->vm_start);
+                  vma->vm_end - vma->vm_start, dd->ipath_unit,
+                  pd->port_port);
 
        if (pgaddr == ureg)
                ret = mmap_ureg(vma, dd, ureg);