Skip to contentSkip to conversation

synclink.c compiler optimiation fix

PATCH In review Paul Fulghum 2 participants 3 messages Sep 7, 2005 – Sep 8, 2005 Char and misc drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 synclink.c compiler optimiation fix

[patch] synclink.c compiler optimization fix

From: Paul Fulghum <[email protected]>

Make some fields of DMA descriptor volatile to
prevent compiler optimizations.

drivers/char/synclink.c +5 -5
@@ -1,7 +1,7 @@
 /*
  * linux/drivers/char/synclink.c
  *
- * $Id: synclink.c,v 4.28 2004/08/11 19:30:01 paulkf Exp $
+ * $Id: synclink.c,v 4.37 2005/09/07 13:13:19 paulkf Exp $
  *
  * Device driver for Microgate SyncLink ISA and PCI
  * high speed multiprotocol serial adapters.
@@ -141,9 +141,9 @@ static MGSL_PARAMS default_params = {
 typedef struct _DMABUFFERENTRY
 {
 	u32 phys_addr;	/* 32-bit flat physical address of data buffer */
-	u16 count;	/* buffer size/data count */
-	u16 status;	/* Control/status field */
-	u16 rcc;	/* character count field */
+	volatile u16 count;	/* buffer size/data count */
+	volatile u16 status;	/* Control/status field */
+	volatile u16 rcc;	/* character count field */
 	u16 reserved;	/* padding required by 16C32 */
 	u32 link;	/* 32-bit flat link to next buffer entry */
 	char *virt_addr;	/* virtual address of data buffer */
@@ -896,7 +896,7 @@ module_param_array(txdmabufs, int, NULL,
 module_param_array(txholdbufs, int, NULL, 0);
 
 static char *driver_name = "SyncLink serial driver";
-static char *driver_version = "$Revision: 4.28 $";
+static char *driver_version = "$Revision: 4.37 $";
 
 static int synclink_init_one (struct pci_dev *dev,
 				     const struct pci_device_id *ent);
#1 lore raw

drivers/char/synclink.c static MGSL_PARAMS default_params = { · line 146

Quoted 6 lines · Paul Fulghum
drivers/char/synclink.c +3 -3
-	u16 count;	/* buffer size/data count */
-	u16 status;	/* Control/status field */
-	u16 rcc;	/* character count field */
+	volatile u16 count;	/* buffer size/data count */
+	volatile u16 status;	/* Control/status field */
+	volatile u16 rcc;	/* character count field */

this is wrong. The structure is in ioremaped memory so you must use reads/writes to access them instead. volatile usage in drivers is never okay - if you are accessing I/O memory you need to use proper acessors, if it is normal memory and you want atomic sematics you need to use the atomic_t type and the operators defined on it.

#2 lore raw
Quoted 2 lines · Christoph Hellwig

The structure is in ioremaped memory so you must
use reads/writes to access them instead.

Yes, using read/write eliminates the compiler optimization and makes the driver portable to other architectures. That change is much more extensive, and it may be a while before I can do a major rewrite of the driver. The volatile change allows the existing driver to work.

Quoted 4 lines

volatile usage in drivers
is never okay - if you are accessing I/O memory you need to use proper acessors, if it is normal memory and you want atomic sematics you need to use the atomic_t type and the operators defined on it.

This is not a matter of atomicity.
It is a matter of hardware DMA causing the value to change without the compiler's knowledge.

If I have a DMA descriptor in normal memory (not the case in the above driver, but it is the case in another driver I maintain) that has fields that do not conform to atomic_t, using volatile seems a valid way of preventing the compiler from optimizing access to the field out of a loop.

Signature · 2 lines
Paul Fulghum
Microgate Systems, Ltd.