Skip to contentSkip to conversation

openposix/fork/7-1.c: A bug fix

PATCH New Bogdan Lezhepekov 2 participants 7 messages Dec 11, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 openposix/fork/7-1.c: A bug fix

The function output interferes with the variable errno, that leads to the false positive result on limited test setups. The issue fixed.

---
 .../open_posix_testsuite/conformance/interfaces/fork/7-1.c    | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +3 -1
index c3db90c00..4249d713d 100644
@@ -53,12 +53,14 @@ static void read_catalog(nl_catd cat, char *who)
 {
 	char *msg = NULL;
 	int i, j;
-	errno = 0;
 
 #if VERBOSE > 0
 	output("Reading the message catalog from %s...\n", who);
 #endif
 
+	/* the output function interferes with errno */ 
+	errno = 0;
+
 	for (i = 1; i <= 2; i++) {
 		for (j = 1; j <= 2; j++) {
 
Signature · 1 line
2.26.2
#1 lore raw

PATCH 1/1 openposix/fork/7-1.c: A bug fix

Hi!

Quoted 17 lines

The function output interferes with the variable errno, that leads to the false positive result on limited test setups. The issue fixed.

---
 .../open_posix_testsuite/conformance/interfaces/fork/7-1.c    | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +0 -1
index c3db90c00..4249d713d 100644
@@ -53,12 +53,14 @@ static void read_catalog(nl_catd cat, char *who)
 {
 	char *msg = NULL;
 	int i, j;
-	errno = 0;

testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c static void read_catalog(nl_catd cat, char *who) · line 65

Quoted 10 lines
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +3 -0
 
 #if VERBOSE > 0
 	output("Reading the message catalog from %s...\n", who);
 #endif
 
+	/* the output function interferes with errno */ 
+	errno = 0;
+
 	for (i = 1; i <= 2; i++) {
 		for (j = 1; j <= 2; j++) {

This is obviously correct, but I would avoid adding the comment, it's kind of obvious that anything that calls to libc may and will interfere with errno.

Also the first line of the commit description could be a bit more description, half of the commits pushed to LTP are bugfixes. So maybe something as:

openposix/fork/7-1.c: Clear errno correctly

...

I can push the patch with these changes if it's okay with you.

Signature · 2 lines
Cyril Hrubis
[email protected]
#2 lore raw

PATCH 1/1 openposix/fork/7-1.c: A bug fix

Hi,

By adding this comment I wanted to stress that the place of "errno = 0" is not randomly chosen, to prevent somebody from moving it back to the beginning of the file :) But if you find it not necessary, then please go ahead.
Thank you,
Bogdan

Quoted 37 lines · Cyril Hrubis

Hi!

Quoted 17 lines

The function output interferes with the variable errno, that leads to the false positive result on limited test setups. The issue fixed.

---
.../open_posix_testsuite/conformance/interfaces/fork/7-1.c | 4 +++-

1 file changed, 3 insertions(+), 1 deletion(-)
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +0 -1
index c3db90c00..4249d713d 100644
@@ -53,12 +53,14 @@ static void read_catalog(nl_catd cat, char *who)
 {
 char *msg = NULL;
 int i, j;
- errno = 0;

testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c static void read_catalog(nl_catd cat, char *who) · line 65

Quoted 10 lines
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +3 -0
 
 #if VERBOSE > 0
 output("Reading the message catalog from %s...\n", who);
 #endif
 
+ /* the output function interferes with errno */
+ errno = 0;
+
 for (i = 1; i <= 2; i++) {
 for (j = 1; j <= 2; j++) {

This is obviously correct, but I would avoid adding the comment, it's kind of obvious that anything that calls to libc may and will interfere with errno.

Also the first line of the commit description could be a bit more description, half of the commits pushed to LTP are bugfixes. So maybe something as:

Quoted 8 lines

openposix/fork/7-1.c: Clear errno correctly
...
I can push the patch with these changes if it's okay with you.

--
Cyril Hrubis
[email protected]

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20201211/ac9a9bb0/attachment.htm>

#3 lore raw

PATCH 1/1 openposix/fork/7-1.c: A bug fix

Hi,

As follow up after chat with Petr Vorel.
A bug in the test was caused by the line "now = localtime(&nw);" in "output" function, as it sets "errno" variable somewhere. -Bogdan

Quoted 49 lines · Bogdan Lezhepekov

Hi,

By adding this comment I wanted to stress that the place of "errno = 0" is not randomly chosen, to prevent somebody from moving it back to the beginning of the file :) But if you find it not necessary, then please go ahead.
Thank you,
Bogdan

Quoted 41 lines · Cyril Hrubis

Hi!

Quoted 17 lines

The function output interferes with the variable errno, that leads to the false positive result on limited test setups. The issue fixed.

---
.../open_posix_testsuite/conformance/interfaces/fork/7-1.c | 4 +++-

1 file changed, 3 insertions(+), 1 deletion(-)
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +0 -1
index c3db90c00..4249d713d 100644
@@ -53,12 +53,14 @@ static void read_catalog(nl_catd cat, char *who)
 {
 char *msg = NULL;
 int i, j;
- errno = 0;

testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c static void read_catalog(nl_catd cat, char *who) · line 65

Quoted 10 lines
testcases/open_posix_testsuite/conformance/interfaces/fork/7-1.c +3 -0
 
 #if VERBOSE > 0
 output("Reading the message catalog from %s...\n", who);
 #endif
 
+ /* the output function interferes with errno */
+ errno = 0;
+
 for (i = 1; i <= 2; i++) {
 for (j = 1; j <= 2; j++) {

This is obviously correct, but I would avoid adding the comment, it's kind of obvious that anything that calls to libc may and will interfere with errno.

Also the first line of the commit description could be a bit more description, half of the commits pushed to LTP are bugfixes. So maybe something as:

openposix/fork/7-1.c: Clear errno correctly
...
I can push the patch with these changes if it's okay with you.

Quoted 8 lines
Quoted 4 lines

--
Cyril Hrubis
[email protected]

--
Mailing list info: https://lists.linux.it/listinfo/ltp

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20201211/a140569e/attachment.htm>

#4 lore raw

PATCH 1/1 openposix/fork/7-1.c: A bug fix

Hi!

Quoted 2 lines

As follow up after chat with Petr Vorel.
A bug in the test was caused by the line "now = localtime(&nw);" in "output" function, as it sets "errno" variable somewhere.

localtime() shouldn't fail with anything but EOVERFLOW it's a bit strange to see it fail there.

Signature · 2 lines
Cyril Hrubis
[email protected]
#5 lore raw

PATCH 1/1 openposix/fork/7-1.c: A bug fix

In reply to Bogdan Lezhepekov

Hi!

Quoted 4 lines

By adding this comment I wanted to stress that the place of "errno = 0" is not randomly chosen, to prevent somebody from moving it back to the beginning of the file :) But if you find it not necessary, then please go ahead.

I think that this still falls under "do not comment the obvious" rule. We do have more than hundred of testcases that do "errno = 0" and as far as I can tell none of them includes a comment that says that we cannot move the code around and I do not think that we should add hundred of comments into these testcases either.

Signature · 2 lines
Cyril Hrubis
[email protected]