Skip to contentSkip to conversation

strstrip incorrectly marked __must_check

PATCH In review James Bottomley 6 participants 12 messages Nov 3, 2009 – Nov 24, 2009

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 strstrip incorrectly marked __must_check

strstrip strips whitespace from the beginning and end of a string. I agree you have to take the returned pointer if you want to strip from the beginning. However, if you wish to keep the whitespace at the beginning and only wish strstrip to remove it from the end, then it's entirely legitimate to discard the returned pointer.

This is what we have in drivers/scsi/ipr.c and the patch to make strstrip __must_check is now causing SCSI spurious warnings in that code.

---

include/linux/string.h +1 -1
index b850886..489019e 100644
@@ -62,7 +62,7 @@ extern char * strnchr(const char *, size_t, int);
 #ifndef __HAVE_ARCH_STRRCHR
 extern char * strrchr(const char *,int);
 #endif
-extern char * __must_check strstrip(char *);
+extern char * strstrip(char *);
 #ifndef __HAVE_ARCH_STRSTR
 extern char * strstr(const char *,const char *);
 #endif
maintainer
#1 lore raw
Quoted 10 lines · On Tue, 03 Nov 2009 12:38:08 -0600 James Bottomley <[email protected]>…

strstrip strips whitespace from the beginning and end of a string. I agree you have to take the returned pointer if you want to strip from the beginning. However, if you wish to keep the whitespace at the beginning and only wish strstrip to remove it from the end, then it's entirely legitimate to discard the returned pointer.

This is what we have in drivers/scsi/ipr.c and the patch to make strstrip __must_check is now causing SCSI spurious warnings in that code.

Would prefer to keep the warning and to patch ipr.c, please. We found I think three call sites which were incorrectly ignoring the strstrip() return value and it's reasonable to fear that others will make the same mistake in the future.

And maybe ipr.c _should_ be patched. Right now it's assuming that the string coming back from the device has no leading whitespace. Why trim any possible trailing whitespace but not trim any possible leading whitespace?

Or..

/*
* Comment goes here
*/
static inline void strsrip_tail(char *str)

{
	char *x __used;
	x = strstrip(str);
}
#2 lore raw

static inline void strsrip_tail(char *str)
{
- char *x __used;
- x = strstrip(str);
+ char *x = strstrip(str);
+ BUG_ON(x != str);
}

Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
#4 lore raw
In reply to Matthew Wilcox

Review

Quoted 7 lines · On Tue, 3 Nov 2009 12:04:20 -0700 Matthew Wilcox <[email protected]> wrote:

static inline void strsrip_tail(char *str)
{

(quoted diff) +2 -2
-	char *x __used;
-	x = strstrip(str);
+	char *x = strstrip(str);
+	BUG_ON(x != str);

That breaks it. It was fine before you did that but now it blows up if there are leading spaces.

#5 lore raw
In reply to Matthew Wilcox

Review

Quoted 7 lines · Matthew Wilcox

static inline void strsrip_tail(char *str)
{

(quoted diff) +2 -2
-	char *x __used;
-	x = strstrip(str);
+	char *x = strstrip(str);
+	BUG_ON(x != str);
 }

Please, no. I said didn't care about leading whitespace ... I didn't say there wasn't any.

James

#6 lore raw
In reply to Andrew Morton
Quoted 14 lines · Andrew Morton
Quoted 10 lines · On Tue, 03 Nov 2009 12:38:08 -0600 James Bottomley <[email protected]>…

strstrip strips whitespace from the beginning and end of a string. I agree you have to take the returned pointer if you want to strip from the beginning. However, if you wish to keep the whitespace at the beginning and only wish strstrip to remove it from the end, then it's entirely legitimate to discard the returned pointer.

This is what we have in drivers/scsi/ipr.c and the patch to make strstrip __must_check is now causing SCSI spurious warnings in that code.

Quoted 4 lines

Would prefer to keep the warning and to patch ipr.c, please. We found I think three call sites which were incorrectly ignoring the strstrip() return value and it's reasonable to fear that others will make the same mistake in the future.

What's the problem with the mistake ... additional leading whitespace?

Quoted 4 lines

And maybe ipr.c _should_ be patched. Right now it's assuming that the string coming back from the device has no leading whitespace. Why trim any possible trailing whitespace but not trim any possible leading whitespace?

I think it doesn't care. It wants to append an error code to the string, and to make it more visible it wants to strip trailing whitespace before doing so.

Quoted 10 lines

Or..

/*
* Comment goes here
*/
static inline void strsrip_tail(char *str)

{
	char *x __used;
	x = strstrip(str);
}

Yes, I could go for that ... I just don't see such a problem with the currently overloaded uses of strstrip.

James

#7 lore raw
In reply to Andrew Morton
Quoted 5 lines

static inline void strsrip_tail(char *str)

{
	char *x __used;
	x = strstrip(str);
}

Bikeshed time but its cleaner to do

static inline __must_check void strstrip(char *str)

{
	return strim(str);
}

and make strim() the old strstrip function without the check requirement

Alan

#8 lore raw

2009/11/4 Alan Cox <[email protected]>:

Quoted 6 lines
Quoted 5 lines

static inline void strsrip_tail(char *str)
{
      char *x __used;
      x = strstrip(str);
}

Quoted 8 lines

Bikeshed time but its cleaner to do

static inline __must_check void strstrip(char *str) {
       return strim(str);
}

and make strim() the old strstrip function without the check requirement

Okey...

[quick hack and compile check]

done :)
sorry for attached file. I'm under poor mail environment now.

From 94a86424a882d52fcd11039d11fbd0363a31472f Mon Sep 17 00:00:00 2001
From: KOSAKI Motohiro <[email protected]>
Date: Wed, 4 Nov 2009 04:47:23 +0900
Subject: [PATCH] lib: Introduce strim

Recently, We marked strstrip() as must_check. because it was frequently misused and it should be checked. however, we found one exception. scsi/ipr.c intentionally ignore return value of strstrip. because, it wish to keep the whitespace at the beginning.

Thus, we need to keep with and without checked whitespace trim function. This patch makes strim new function and ipr.c use it.

---
 drivers/scsi/ipr.c     |    4 ++--
 include/linux/string.h |    9 ++++++++-
 lib/string.c           |    6 +++---
 3 files changed, 13 insertions(+), 6 deletions(-)
drivers/scsi/ipr.c +2 -2
index d40d5c7..6602fc4 100644
@@ -1333,7 +1333,7 @@ static void ipr_log_enhanced_dual_ioa_error(struct ipr_ioa_cfg *ioa_cfg,
 
 	error = &hostrcb->hcam.u.error.u.type_17_error;
 	error->failure_reason[sizeof(error->failure_reason) - 1] = '\0';
-	strstrip(error->failure_reason);
+	strim(error->failure_reason);
 
 	ipr_hcam_err(hostrcb, "%s [PRC: %08X]\n", error->failure_reason,
 		     be32_to_cpu(hostrcb->hcam.u.error.prc));
@@ -1359,7 +1359,7 @@ static void ipr_log_dual_ioa_error(struct ipr_ioa_cfg *ioa_cfg,
 
 	error = &hostrcb->hcam.u.error.u.type_07_error;
 	error->failure_reason[sizeof(error->failure_reason) - 1] = '\0';
-	strstrip(error->failure_reason);
+	strim(error->failure_reason);
 
 	ipr_hcam_err(hostrcb, "%s [PRC: %08X]\n", error->failure_reason,
 		     be32_to_cpu(hostrcb->hcam.u.error.prc));
include/linux/string.h +8 -1
index b850886..17a375e 100644
@@ -62,7 +62,14 @@ extern char * strnchr(const char *, size_t, int);
 #ifndef __HAVE_ARCH_STRRCHR
 extern char * strrchr(const char *,int);
 #endif
-extern char * __must_check strstrip(char *);
+
+extern char * strim(char *);
+
+static inline __must_check char* strstrip(char *str)
+{
+	return strim(str);
+}
+
 #ifndef __HAVE_ARCH_STRSTR
 extern char * strstr(const char *,const char *);
 #endif
lib/string.c +3 -3
index b19b87a..28211c6 100644
@@ -330,14 +330,14 @@ EXPORT_SYMBOL(strnchr);
 #endif
 
 /**
- * strstrip - Removes leading and trailing whitespace from @s.
+ * strim - Removes leading and trailing whitespace from @s.
  * @s: The string to be stripped.
  *
  * Note that the first trailing whitespace is replaced with a %NUL-terminator
  * in the given string @s. Returns a pointer to the first non-whitespace
  * character in @s.
  */
-char *strstrip(char *s)
+char *strim(char *s)
 {
 	size_t size;
 	char *end;
@@ -357,7 +357,7 @@ char *strstrip(char *s)
 
 	return s;
 }
-EXPORT_SYMBOL(strstrip);
+EXPORT_SYMBOL(strim);
 
 #ifndef __HAVE_ARCH_STRLEN
 /**
Signature · 1 line
1.6.2.5
#9 lore raw

Hi,

I have several places in my code where the new __must_check of strstrip will introduce unnecessary dummy variables to avoid the warnings.

Therefore I would like to have the suggested new strim() or strstip_tail() function. Any chance to have this upstream soon?

Michael

Quoted 22 lines · KOSAKI Motohiro

2009/11/4 Alan Cox <[email protected]>:

Quoted 6 lines
Quoted 5 lines

static inline void strsrip_tail(char *str)

{
      char *x __used;
      x = strstrip(str);
}
Quoted 8 lines

Bikeshed time but its cleaner to do

static inline __must_check void strstrip(char *str)

{
       return strim(str);
}

and make strim() the old strstrip function without the check requirement

Okey...

[quick hack and compile check]

done :)
sorry for attached file. I'm under poor mail environment now.

#10 lore raw
Quoted 7 lines

Hi,

I have several places in my code where the new __must_check of strstrip will introduce unnecessary dummy variables to avoid the warnings.

Therefore I would like to have the suggested new strim() or strstip_tail() function. Any chance to have this upstream soon?

strim() is in mmotm now. I expect it will merge mainline soon.

Thanks.

Quoted 27 lines

Michael

Quoted 22 lines · KOSAKI Motohiro

2009/11/4 Alan Cox <[email protected]>:

Quoted 6 lines
Quoted 5 lines

static inline void strsrip_tail(char *str)

{
      char *x __used;
      x = strstrip(str);
}
Quoted 8 lines

Bikeshed time but its cleaner to do

static inline __must_check void strstrip(char *str)

{
       return strim(str);
}

and make strim() the old strstrip function without the check requirement

Okey...

[quick hack and compile check]

done :)
sorry for attached file. I'm under poor mail environment now.

#11 lore raw
Quoted 9 lines · KOSAKI Motohiro
Quoted 7 lines

Hi,

I have several places in my code where the new __must_check of strstrip will introduce unnecessary dummy variables to avoid the warnings.

Therefore I would like to have the suggested new strim() or strstip_tail() function. Any chance to have this upstream soon?

strim() is in mmotm now. I expect it will merge mainline soon.

That's really good news!

Thanks!

Michael