dcdbas: force SMI to happen when expected

PATCH Applied stuart h 3 participants 4 messages Feb 25, 2011 – Mar 2, 2011

Share & tools

stuart h

author

2011-02-25 02:41 UTC #0 lore raw

PATCH v2 1/1 dcdbas: force SMI to happen when expected

This is a reworked version of a very similar patch I sent about three weeks ago... a flaw was pointed out with the original. That patch didn't make it into the kernel, though, so I'm just sending this one as a new patch.

The dcdbas driver can do an I/O write to cause a SMI to occur. The SMI handler looks at certain registers and memory locations, so the SMI needs to happen immediately. On some systems I/O writes are posted, though, causing the SMI to happen well after the "outb" occurred, which causes random failures. Following the "outb" with an "inb" forces the write to go through even if it is posted.

This patch is against the 2.6.37.1 kernel.

Signed-off-by Stuart Hayes <[email protected]>

---

drivers/firmware/dcdbas.c +3 -1

@@ -268,8 +268,10 @@ int dcdbas_smi_request(struct smi_cmd *s
 	}
 
 	/* generate SMI */
+	/* inb to force posted write through and make SMI happen now */
 	asm volatile (
-		"outb %b0,%w1"
+		"outb %b0,%w1\n"
+		"inb %w1"
 		: /* no output args */
 		: "a" (smi_cmd->command_code),
 		  "d" (smi_cmd->command_address),

Jiri Kosina

2011-03-01 14:57 UTC #1 lore raw

Quoted 12 lines · stuart h

This is a reworked version of a very similar patch I sent about three weeks ago... a flaw was pointed out with the original. That patch didn't make it into the kernel, though, so I'm just sending this one as a new patch.
The dcdbas driver can do an I/O write to cause a SMI to occur. The SMI handler looks at certain registers and memory locations, so the SMI needs to happen immediately. On some systems I/O writes are posted, though, causing the SMI to happen well after the "outb" occurred, which causes random failures. Following the "outb" with an "inb" forces the write to go through even if it is posted.

Quoted 3 lines

This patch is against the 2.6.37.1 kernel.
Signed-off-by Stuart Hayes <[email protected]>

As this doesn't seem to be picked up ... Douglas, can I have your
Acked-by: please, so that I could merge it (as Acked by maintainer?).

Thanks.

Quoted 17 lines

drivers/firmware/dcdbas.c +3 -1

@@ -268,8 +268,10 @@ int dcdbas_smi_request(struct smi_cmd *s
 	}
 
 	/* generate SMI */
+	/* inb to force posted write through and make SMI happen now */
 	asm volatile (
-		"outb %b0,%w1"
+		"outb %b0,%w1\n"
+		"inb %w1"
 		: /* no output args */
 		: "a" (smi_cmd->command_code),
 		  "d" (smi_cmd->command_address),

Signature · 2 lines

Jiri Kosina
SUSE Labs, Novell Inc.

[email protected]

Acked-by on v2 1/1

2011-03-01 15:54 UTC #2 lore raw

In reply to stuart h

Quoted 24 lines · Stuart Hayes

This is a reworked version of a very similar patch I sent about three weeks ago... a flaw was pointed out with the original. That patch didn't make it into the kernel, though, so I'm just sending this one as a new patch.
The dcdbas driver can do an I/O write to cause a SMI to occur. The SMI handler looks at certain registers and memory locations, so the SMI needs to happen immediately. On some systems I/O writes are posted, though, causing the SMI to happen well after the "outb" occurred, which causes random failures. Following the "outb" with an "inb" forces the write to go through even if it is posted.
This patch is against the 2.6.37.1 kernel.
Signed-off-by Stuart Hayes <[email protected]>
---
--- linux-2.6.37/drivers/firmware/dcdbas.c 2011-01-18
10:44:44.869882461 -0500
+++ linux-2.6.37_test/drivers/firmware/dcdbas.c 2011-02-24
13:23:58.663771596 -0500
drivers/firmware/dcdbas.c +0 -0
@@ -268,8 +268,10 @@ int dcdbas_smi_request(struct smi_cmd *s
 	}

drivers/firmware/dcdbas.c int dcdbas_smi_request(struct smi_cmd *s · line 277

Quoted 10 lines

drivers/firmware/dcdbas.c +3 -1

 
 	/* generate SMI */
+	/* inb to force posted write through and make SMI happen now */
 	asm volatile (
-		"outb %b0,%w1"
+		"outb %b0,%w1\n"
+		"inb %w1"
 		: /* no output args */
 		: "a" (smi_cmd->command_code),
 		  "d" (smi_cmd->command_address),

Acked-by Doug Warzecha <[email protected]>

Jiri Kosina

Applied

2011-03-02 12:42 UTC #3 lore raw

Quoted 24 lines · On Tue, 1 Mar 2011, [email protected] wrote:

Quoted 24 lines
This is a reworked version of a very similar patch I sent about three weeks ago... a flaw was pointed out with the original. That patch didn't make it into the kernel, though, so I'm just sending this one as a new patch.
The dcdbas driver can do an I/O write to cause a SMI to occur. The SMI handler looks at certain registers and memory locations, so the SMI needs to happen immediately. On some systems I/O writes are posted, though, causing the SMI to happen well after the "outb" occurred, which causes random failures. Following the "outb" with an "inb" forces the write to go through even if it is posted.
This patch is against the 2.6.37.1 kernel.
Signed-off-by Stuart Hayes <[email protected]>
---
--- linux-2.6.37/drivers/firmware/dcdbas.c 2011-01-18
10:44:44.869882461 -0500
+++ linux-2.6.37_test/drivers/firmware/dcdbas.c 2011-02-24
13:23:58.663771596 -0500
(quoted diff) +0 -0
@@ -268,8 +268,10 @@ int dcdbas_smi_request(struct smi_cmd *s
 	}
 

Quoted 12 lines

drivers/firmware/dcdbas.c int dcdbas_smi_request(struct smi_cmd *s · line 277
Quoted 10 lines
drivers/firmware/dcdbas.c +3 -1
 	/* generate SMI */
+	/* inb to force posted write through and make SMI happen now */
 	asm volatile (
-		"outb %b0,%w1"
+		"outb %b0,%w1\n"
+		"inb %w1"
 		: /* no output args */
 		: "a" (smi_cmd->command_code),
 		  "d" (smi_cmd->command_address),
Acked-by Doug Warzecha <[email protected]>

Applied, thanks.

Signature · 2 lines

Jiri Kosina
SUSE Labs, Novell Inc.