Skip to contentSkip to conversation

scsi: fix bad use of udelay in atp870u.c

PATCH In review [email protected] 5 participants 7 messages Jan 9, 2009 – Apr 24, 2012 SCSI

lore
Share & tools
lore
#0 lore raw

PATCH 10/11 scsi: fix bad use of udelay in atp870u.c

From: Martin Michlmayr <[email protected]>

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1
Fix this by using a combination of mdelay and udelay.

---

 drivers/scsi/atp870u.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/scsi/atp870u.c renamed from drivers/scsi/atp870u.c~scsi-fix-bad-use-of-udelay-in-atp870uc +2 -1
diff -puN drivers/scsi/atp870u.c~scsi-fix-bad-use-of-udelay-in-atp870uc drivers/scsi/atp870u.c
@@ -1171,7 +1171,8 @@ wait_io1:
 	outw(val, tmport);
 	outb(2, 0x80);
 TCM_SYNC:
-	udelay(0x800);
+	mdelay(2);
+	udelay(48);
 	if ((inb(tmport) & 0x80) == 0x00) {	/* bsy ? */
 		outw(0, tmport--);
 		outb(0, tmport);

_

#2 lore raw
In reply to Martin Michlmayr
Quoted 8 lines · On Fri, 9 Jan 2009 21:58:35 +0100 Martin Michlmayr <[email protected]> wrote:
Quoted 5 lines · * [email protected] <[email protected]> [2009-01-09 12:28]:

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1
Fix this by using a combination of mdelay and udelay.

James already said that this patch is wrong,

Well, it's not "wrong" - it has no runtime effect - it just fixes the build.

Quoted 3 lines

so I suggest you drop
this patch. (Even if this will probably mean that this issue will never get addressed.)

shrug.

maintainer
#3 lore raw
Quoted 10 lines · Andrew Morton
Quoted 7 lines · On Fri, 9 Jan 2009 21:58:35 +0100 Martin Michlmayr <[email protected]> wrote:
Quoted 5 lines · * [email protected] <[email protected]> [2009-01-09 12:28]:

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1
Fix this by using a combination of mdelay and udelay.

Quoted 3 lines
Quoted 1 line

James already said that this patch is wrong,

Well, it's not "wrong" - it has no runtime effect - it just fixes the build.

It's wrong to silence a warning or build break while keeping the effect it was complaining about it's hiding a bug. Now if the warning is wrong, we can take it out of the ARM build ... but I've got to say it looks right: the udelay in this driver will lock a UP system solid for 2ms.

James

#1 lore raw
In reply to [email protected]
Quoted 5 lines · * [email protected] <[email protected]> [2009-01-09 12:28]:

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1
Fix this by using a combination of mdelay and udelay.

James already said that this patch is wrong, so I suggest you drop this patch. (Even if this will probably mean that this issue will never get addressed.)

Signature · 2 lines
Martin Michlmayr
http://www.cyrius.com/
#4 lore raw
In reply to James Bottomley
Quoted 5 lines · James Bottomley

It's wrong to silence a warning or build break while keeping the effect it was complaining about it's hiding a bug. Now if the warning is wrong, we can take it out of the ARM build ... but I've got to say it looks right: the udelay in this driver will lock a UP system solid for 2ms.

mdelay exist for a reason. Even if this code is really bad adding a comment about that and moving on is better than leaving the too large udelay there.

#5 lore raw

PATCH v2 1/1 atp870u: Fix bad use of udelay

In reply to James Bottomley

From: Martin Michlmayr <[email protected]>

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1

This is because udelay is defined on ARM, roughly speaking, as

	#define udelay(n) ((n) > 2000 ? __bad_udelay() : \
		__const_udelay((n) * ((2199023U*HZ)>>11)))

The argument to __const_udelay is the number of jiffies to wait divided by 4, but this does not work unless the multiplication does not overflow, and that is what the build error is designed to prevent. The intended behavior can be achieved by using mdelay to call udelay multiple times in a loop.

---
Hi James,

Three years ago, you wrote[1]:

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines · akpm at linux-foundation.org
Quoted 6 lines

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1
Fix this by using a combination of mdelay and udelay.

[...]

Quoted 5 lines

It's wrong to silence a warning or build break while keeping the effect it was complaining about it's hiding a bug. Now if the warning is wrong, we can take it out of the ARM build ... but I've got to say it looks right: the udelay in this driver will lock a UP system solid for 2ms.

Sorry for the very slow response. I think the patch was inadequately explained and that it is actually a good patch.

Here's the patch again with a description that helped me convince myself it is ok. Could you look it over and let me know what you think?

Thanks,
Jonathan

[1] http://thread.gmane.org/gmane.linux.scsi/47523/focus=47533

 drivers/scsi/atp870u.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/scsi/atp870u.c +10 -1
index 68ce08552f69..a540162ac59c 100644
@@ -1173,7 +1173,16 @@ wait_io1:
 	outw(val, tmport);
 	outb(2, 0x80);
 TCM_SYNC:
-	udelay(0x800);
+	/*
+	 * The funny division into multiple delays is to accomodate
+	 * arches like ARM where udelay() multiplies its argument by
+	 * a large number to initialize a loop counter.  To avoid
+	 * overflow, the maximum supported udelay is 2000 microseconds.
+	 *
+	 * XXX it would be more polite to find a way to use msleep()
+	 */
+	mdelay(2);
+	udelay(48);
 	if ((inb(tmport) & 0x80) == 0x00) {	/* bsy ? */
 		outw(0, tmport--);
 		outb(0, tmport);
Signature · 1 line
1.7.10
#6 lore raw

PATCH v2 1/1 atp870u: Fix bad use of udelay

Quoted 19 lines · On Tue, 24 Apr 2012 02:49:16 -0500 Jonathan Nieder <[email protected]> wrote:

From: Martin Michlmayr <[email protected]>

The ACARD driver calls udelay() with a value > 2000, which leads to to the following compilation error on ARM:
ERROR: "__bad_udelay" [drivers/scsi/atp870u.ko] undefined! make[1]: *** [__modpost] Error 1

...

drivers/scsi/atp870u.c +3 -1
@@ -1173,7 +1173,16 @@ wait_io1:
 	outw(val, tmport);
 	outb(2, 0x80);
 TCM_SYNC:
-	udelay(0x800);
+	/*
+	 * The funny division into multiple delays is to accomodate
+	 * arches like ARM where udelay() multiplies its argument by

drivers/scsi/atp870u.c wait_io1: · line 1188

Quoted 10 lines
drivers/scsi/atp870u.c +7 -0
+	 * a large number to initialize a loop counter.  To avoid
+	 * overflow, the maximum supported udelay is 2000 microseconds.
+	 *
+	 * XXX it would be more polite to find a way to use msleep()
+	 */
+	mdelay(2);
+	udelay(48);
 	if ((inb(tmport) & 0x80) == 0x00) {	/* bsy ? */
 		outw(0, tmport--);
 		outb(0, tmport);

Fair enough, I say. Looking at this driver, I do think it's best to minimally fix the build and then tiptoe away very delicately lest something explode.

That being said, I'm glad the kernel has a function called "fun_scam".