fix readv01 testcase for parisc architecture

From: "Helge Deller" <[email protected]>
To: [email protected]
Sent: Tuesday, 19 November, 2013 10:46:12 PM
Subject: [LTP] [PATCH] fix readv01 testcase for parisc architecture

In the second test of readv01 the rd_iovec[] array was not defined or zero-initialized for vectors which will be accessed by the readv() syscall. This leads to the fact that on the parisc platform the readv syscall accessed random memory which is located behind the array. Fix this problem by adding two more vectors with address NULL and length 0.

Furthermore, in the test2 itself we tell readv() to read 4 vecors, while the error message behind it says that it has read CHUNK bytes "followed by *two* NULL vectors", which then sums up to a total of 3 vectors instead of 4.
Fix this by calling readv() with only 3 vectors instead of 4.

Signed-off-by Helge Deller <[email protected]>

testcases/kernel/syscalls/readv/readv01.c +4 -3

index d92c1be..ca83310 100644
@@ -66,7 +66,9 @@ struct iovec rd_iovec[MAX_IOVEC] = {
 	{(buf2 + CHUNK * 10), CHUNK},
 
 	/* Test case #2 */
-	{(buf2 + CHUNK * 11), CHUNK}
+	{(buf2 + CHUNK * 11), CHUNK},
+	{NULL, 0},
+	{NULL, 0}
 };
 
 char f_name[K_1];

@@ -112,7 +114,7 @@ int main(int ac, char **av)
 
 //test2:
 		l_seek(fd, CHUNK * 12, 0);
-		if (readv(fd, (rd_iovec + 1), 4) != CHUNK) {
+		if (readv(fd, (rd_iovec + 1), 3) != CHUNK) {
 			tst_resm(TFAIL, "readv failed reading %d bytes, "
 				 "followed by two NULL vectors", CHUNK);
 		} else {
 
------------------------------------------------------------------------------

Shape the Mobile Experience: Free Subscription

Software experts and developers: Be at the forefront of tech innovation. Intel(R) Software Adrenaline delivers strategic insight and game-changing conversations that shape the rapidly evolving mobile landscape. Sign up now.
http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

Quoted 9 lines

From: "Helge Deller" <[email protected]>
Sent: Tuesday, 19 November, 2013 10:46:12 PM
Subject: [LTP] [PATCH] fix readv01 testcase for parisc architecture

In the second test of readv01 the rd_iovec[] array was not defined or zero-initialized for vectors which will be accessed by the readv() syscall. This leads to the fact that on the parisc platform the readv syscall accessed random memory which is located behind the array. Fix this problem by adding two more vectors with address NULL and length 0.

Hi,

It shouldn't be behind, since rd_iovec array has size of MAX_IOVEC == 16.

It's partially initialized, and looking at C99 6.7.8.21, I'd expect the
rest to be zero-initialized as well:
"If there are fewer initializers in a brace-enclosed list than there are elements or members of an aggregate, or fewer characters in a string literal used to initialize an array of known size than there are elements in the array, the remainder of the aggregate shall be initialized implicitly the same as objects that have static storage duration."

If you add 'static' to rd_iovec array, does the problem go away? What compiler are you using?

Quoted 5 lines

Furthermore, in the test2 itself we tell readv() to read 4 vecors, while the error message behind it says that it has read CHUNK bytes "followed by *two* NULL vectors", which then sums up to a total of 3 vectors instead of 4.
Fix this by calling readv() with only 3 vectors instead of 4.

Agreed.

Quoted 20 lines

Signed-off-by Helge Deller <[email protected]>

testcases/kernel/syscalls/readv/readv01.c +3 -1

index d92c1be..ca83310 100644
@@ -66,7 +66,9 @@ struct iovec rd_iovec[MAX_IOVEC] = {
 	{(buf2 + CHUNK * 10), CHUNK},
 
 	/* Test case #2 */
-	{(buf2 + CHUNK * 11), CHUNK}
+	{(buf2 + CHUNK * 11), CHUNK},
+	{NULL, 0},
+	{NULL, 0}
 };
 
 char f_name[K_1];

@@ -112,7 +114,7 @@ int main(int ac, char **av)
 

testcases/kernel/syscalls/readv/readv01.c int main(int ac, char **av) · line 120

Quoted 8 lines

testcases/kernel/syscalls/readv/readv01.c +1 -1

 //test2:
 		l_seek(fd, CHUNK * 12, 0);
-		if (readv(fd, (rd_iovec + 1), 4) != CHUNK) {
+		if (readv(fd, (rd_iovec + 1), 3) != CHUNK) {
 			tst_resm(TFAIL, "readv failed reading %d bytes, "
 				 "followed by two NULL vectors", CHUNK);
 		} else {

Ok, so if you could just apply this second hunk it would be great!

Thanks for maintaining LTP!
It helped me a lot to find bugs in the parisc arch code of the kernel.