On Fri, 19 May 2006 11:15:27 -0500 James Bottomley wrote:
> On Fri, 2006-05-19 at 09:08 -0700, Randy.Dunlap wrote:
> > + .unchecked_isa_dma = 1, /* ppa cannot deal with highmem, so
> > + * this is an easy trick to ensure
> > + * all io pages for this host reside
> > + * in low memory */
>
> You still don't need this piece, do you? It will cause all the kmallocs
> still to be GFP_DMA ... allocations from zone normal should be fine,
> shouldn't they?
darn. Looks correct and stress testing didn't cause it to fail.
Thanks.
Sounds like we should kill the .unchecked_isa_dma = 1 in the imm driver
also, to prevent ISA (16 MB) pressure.
---
From: Randy Dunlap <rdunlap@xxxxxxxxxxxx>
ppa cannot handle highmem pages, and like imm, which already has
this patch, the device is slow, so performance is not a big issue,
so just force pages to be in low memory (hence mapped).
Signed-off-by: Randy Dunlap <rdunlap@xxxxxxxxxxxx>
---
drivers/scsi/ppa.c | 7 +++++++
1 files changed, 7 insertions(+)
--- linux-2.6.17-rc4.orig/drivers/scsi/ppa.c
+++ linux-2.6.17-rc4/drivers/scsi/ppa.c
@@ -982,6 +982,12 @@ static int device_check(ppa_struct *dev)
return -ENODEV;
}
+static int ppa_adjust_queue(struct scsi_device *device)
+{
+ blk_queue_bounce_limit(device->request_queue, BLK_BOUNCE_HIGH);
+ return 0;
+}
+
static struct scsi_host_template ppa_template = {
.module = THIS_MODULE,
.proc_name = "ppa",
@@ -997,6 +1003,7 @@ static struct scsi_host_template ppa_tem
.cmd_per_lun = 1,
.use_clustering = ENABLE_CLUSTERING,
.can_queue = 1,
+ .slave_alloc = ppa_adjust_queue,
};
/***************************************************************************
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
