Skip to contentSkip to conversation

m68k: fix PIO case in esp

PATCH In review Al Viro 1 participant 1 message Jan 3, 2006 SCSI

lore
Share & tools
lore

Al Viro

author
#0 lore raw

PATCH 40/41 m68k: fix PIO case in esp

From: Al Viro <[email protected]>
Date: 1135537383 -0500

we always set ->SCp.ptr to physical address of buffer; for DMA that's just what we need, but we end up using it as virtual address in PIO case of esp_do_data(), with obvious breakage as soon as memory mapping becomes non-trivial. The fix is obvious.

---

 drivers/scsi/NCR53C9x.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)

7f89e0aea371b83e79fb3752b7dd6deeb21f01d9

drivers/scsi/NCR53C9x.c +3 -2
index 640590b..c7dd015 100644
@@ -1799,6 +1799,7 @@ static int esp_do_data(struct NCR_ESP *e
 		 */
 		int oldphase, i = 0; /* or where we left off last time ?? esp->current_data ?? */
 		int fifocnt = 0;
+		unsigned char *p = phys_to_virt((unsigned long)SCptr->SCp.ptr);
 
 		oldphase = esp_read(eregs->esp_status) & ESP_STAT_PMASK;
 
@@ -1860,7 +1861,7 @@ static int esp_do_data(struct NCR_ESP *e
 
 				/* read fifo */
 				for(j=0;j<fifocnt;j++)
-					SCptr->SCp.ptr[i++] = esp_read(eregs->esp_fdata);
+					p[i++] = esp_read(eregs->esp_fdata);
 
 				ESPDATA(("(%d) ", i));
 
@@ -1882,7 +1883,7 @@ static int esp_do_data(struct NCR_ESP *e
 
 				/* fill fifo */
 				for(j=0;j<this_count;j++)
-					esp_write(eregs->esp_fdata, SCptr->SCp.ptr[i++]);
+					esp_write(eregs->esp_fdata, p[i++]);
 
 				/* how many left if this goes out ?? */
 				hmuch -= this_count;
Signature · 1 line
0.99.9.GIT