Skip to contentSkip to conversation

tests/boot-serial-test: Fix problem with timeout due to dropped characters

PATCH Applied Thomas Huth 3 participants 6 messages Feb 16, 2018 – Feb 21, 2018

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 tests/boot-serial-test: Fix problem with timeout due to dropped characters

Commit 92b540dac9fc3a5 introduce a counter to handle the timeouts in a better way. But in case ccnt reaches 512, the current read character is ignored - and if that character is part of the string that we are looking for, the test fails to match the string.

Almost all of the tests look for a string within the first 512 bytes of firmware output, so the problem never triggered there. But the hppa test that has been added recently looks for a longer string at the very end of a long output, thus there's a chance that we miss a character there so that the test fails unexpectedly. Fix it by *not* reading and dropping a character if the counter reaches 512.

Fixes 92b540dac9fc3a572c7342edd0b073000f5a6abf

---
@Peter: Since this fixes the problem with running "make check", could you maybe apply this directly to the master branch? Thanks, and sorry for the inconvenience!

 tests/boot-serial-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/boot-serial-test.c +1 -1
index ea87a80..696f7a3 100644
@@ -101,7 +101,7 @@ static void check_guest_output(const testdef_t *test, int fd)
     /* Poll serial output... Wait at most 60 seconds */
     for (i = 0; i < 6000; ++i) {
         ccnt = 0;
-        while ((nbr = read(fd, &ch, 1)) == 1 && ccnt++ < 512) {
+        while (ccnt++ < 512 && (nbr = read(fd, &ch, 1)) == 1) {
             if (ch == test->expect[pos]) {
                 pos += 1;
                 if (test->expect[pos] == '\0') {
Signature · 1 line
1.8.3.1
#1 lore raw
Quoted 27 lines · Thomas Huth

Commit 92b540dac9fc3a5 introduce a counter to handle the timeouts in a better way. But in case ccnt reaches 512, the current read character is ignored - and if that character is part of the string that we are looking for, the test fails to match the string.

Almost all of the tests look for a string within the first 512 bytes of firmware output, so the problem never triggered there. But the hppa test that has been added recently looks for a longer string at the very end of a long output, thus there's a chance that we miss a character there so that the test fails unexpectedly. Fix it by *not* reading and dropping a character if the counter reaches 512.

Fixes 92b540dac9fc3a572c7342edd0b073000f5a6abf

---
@Peter: Since this fixes the problem with running "make check", could you maybe apply this directly to the master branch? Thanks, and sorry for the inconvenience!

 tests/boot-serial-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/boot-serial-test.c +0 -0
index ea87a80..696f7a3 100644

tests/boot-serial-test.c static void check_guest_output(const testdef_t *test, int fd) · line 107

Quoted 8 lines
tests/boot-serial-test.c +1 -1
@@ -101,7 +101,7 @@ static void check_guest_output(const testdef_t *test, int fd)
     /* Poll serial output... Wait at most 60 seconds */
     for (i = 0; i < 6000; ++i) {
         ccnt = 0;
-        while ((nbr = read(fd, &ch, 1)) == 1 && ccnt++ < 512) {
+        while (ccnt++ < 512 && (nbr = read(fd, &ch, 1)) == 1) {
             if (ch == test->expect[pos]) {
                 pos += 1;
                 if (test->expect[pos] == '\0') {

I did a test build with this, but OpenBSD's compiler now complains:

/home/qemu/tests/boot-serial-test.c: In function 'test_machine':
/usr/local/include/glib-2.0/glib/gmacros.h:370:7: warning: 'nbr' may be used uninitialized in this function [-Wmaybe-uninitialized]

    if (expr)                                    \
       ^
/home/qemu/tests/boot-serial-test.c:114:12: note: 'nbr' was declared here
     int i, nbr, pos = 0, ccnt;
            ^

This is obviously a false positive, but we can silence it by

tests/boot-serial-test.c +1 -1
@@ -111,7 +111,7 @@ static testdef_t tests[] = {
 static void check_guest_output(const testdef_t *test, int fd)
 {
     bool output_ok = false;
-    int i, nbr, pos = 0, ccnt;
+    int i, nbr = 0, pos = 0, ccnt;
     char ch;
 
     /* Poll serial output... Wait at most 60 seconds */

so I'll just squash that change in if that's ok?

thanks
-- PMM

#2 lore raw
Quoted 62 lines · Peter Maydell
Quoted 27 lines · Thomas Huth

Commit 92b540dac9fc3a5 introduce a counter to handle the timeouts in a better way. But in case ccnt reaches 512, the current read character is ignored - and if that character is part of the string that we are looking for, the test fails to match the string.

Almost all of the tests look for a string within the first 512 bytes of firmware output, so the problem never triggered there. But the hppa test that has been added recently looks for a longer string at the very end of a long output, thus there's a chance that we miss a character there so that the test fails unexpectedly. Fix it by *not* reading and dropping a character if the counter reaches 512.

Fixes 92b540dac9fc3a572c7342edd0b073000f5a6abf

---
@Peter: Since this fixes the problem with running "make check", could you maybe apply this directly to the master branch? Thanks, and sorry for the inconvenience!

 tests/boot-serial-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/boot-serial-test.c +0 -0
index ea87a80..696f7a3 100644

tests/boot-serial-test.c static void check_guest_output(const testdef_t *test, int fd) · line 107

Quoted 8 lines
tests/boot-serial-test.c +1 -1
@@ -101,7 +101,7 @@ static void check_guest_output(const testdef_t *test, int fd)
     /* Poll serial output... Wait at most 60 seconds */
     for (i = 0; i < 6000; ++i) {
         ccnt = 0;
-        while ((nbr = read(fd, &ch, 1)) == 1 && ccnt++ < 512) {
+        while (ccnt++ < 512 && (nbr = read(fd, &ch, 1)) == 1) {
             if (ch == test->expect[pos]) {
                 pos += 1;
                 if (test->expect[pos] == '\0') {

I did a test build with this, but OpenBSD's compiler now complains:

/home/qemu/tests/boot-serial-test.c: In function 'test_machine':
/usr/local/include/glib-2.0/glib/gmacros.h:370:7: warning: 'nbr' may be used uninitialized in this function [-Wmaybe-uninitialized]

    if (expr)                                    \
       ^
/home/qemu/tests/boot-serial-test.c:114:12: note: 'nbr' was declared here
     int i, nbr, pos = 0, ccnt;
            ^

This is obviously a false positive, but we can silence it by

tests/boot-serial-test.c +1 -1
@@ -111,7 +111,7 @@ static testdef_t tests[] = {
 static void check_guest_output(const testdef_t *test, int fd)
 {
     bool output_ok = false;
-    int i, nbr, pos = 0, ccnt;
+    int i, nbr = 0, pos = 0, ccnt;
     char ch;
 
     /* Poll serial output... Wait at most 60 seconds */
Quoted 1 line

so I'll just squash that change in if that's ok?

Sure!

Thanks,
Thomas

PS: I think the OpenBSD compiler is wrong here, nbr should get initialized at least once before the g_assert(nbr >= 0) check ...

#3 lore raw
Quoted 2 lines · Thomas Huth

PS: I think the OpenBSD compiler is wrong here, nbr should get initialized at least once before the g_assert(nbr >= 0) check ...

Yes, I agree, but it does require some intelligence on the part of the compiler to figure out that the ccnt++ < 512 condition can't trigger first time round the loop. (It's a gcc 4.9.3, so presumably newer gcc are indeed smarter.)

Applied to master.

thanks
-- PMM

#4 lore raw
In reply to Thomas Huth
Quoted 28 lines · Thomas Huth

Commit 92b540dac9fc3a5 introduce a counter to handle the timeouts in a better way. But in case ccnt reaches 512, the current read character is ignored - and if that character is part of the string that we are looking for, the test fails to match the string.

Almost all of the tests look for a string within the first 512 bytes of firmware output, so the problem never triggered there. But the hppa test that has been added recently looks for a longer string at the very end of a long output, thus there's a chance that we miss a character there so that the test fails unexpectedly. Fix it by *not* reading and dropping a character if the counter reaches 512.

Fixes 92b540dac9fc3a572c7342edd0b073000f5a6abf

---
@Peter: Since this fixes the problem with running "make check", could you maybe apply this directly to the master branch? Thanks, and sorry for the inconvenience!

 tests/boot-serial-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tests/boot-serial-test.c +0 -0
index ea87a80..696f7a3 100644

tests/boot-serial-test.c static void check_guest_output(const testdef_t *test, int fd) · line 107

Quoted 8 lines
tests/boot-serial-test.c +1 -1
@@ -101,7 +101,7 @@ static void check_guest_output(const testdef_t *test, int fd)
     /* Poll serial output... Wait at most 60 seconds */
     for (i = 0; i < 6000; ++i) {
         ccnt = 0;
-        while ((nbr = read(fd, &ch, 1)) == 1 && ccnt++ < 512) {
+        while (ccnt++ < 512 && (nbr = read(fd, &ch, 1)) == 1) {
             if (ch == test->expect[pos]) {
                 pos += 1;
                 if (test->expect[pos] == '\0') {

Queued in the meanwhile, thanks.

Paolo

#5 lore raw
Quoted 23 lines · Paolo Bonzini
Quoted 22 lines · Thomas Huth

Commit 92b540dac9fc3a5 introduce a counter to handle the timeouts in a better way. But in case ccnt reaches 512, the current read character is ignored - and if that character is part of the string that we are looking for, the test fails to match the string.

Almost all of the tests look for a string within the first 512 bytes of firmware output, so the problem never triggered there. But the hppa test that has been added recently looks for a longer string at the very end of a long output, thus there's a chance that we miss a character there so that the test fails unexpectedly. Fix it by *not* reading and dropping a character if the counter reaches 512.

Fixes 92b540dac9fc3a572c7342edd0b073000f5a6abf

---
@Peter: Since this fixes the problem with running "make check", could you maybe apply this directly to the master branch? Thanks, and sorry for the inconvenience!

 tests/boot-serial-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines
Quoted 6 lines
tests/boot-serial-test.c +0 -0
index ea87a80..696f7a3 100644

tests/boot-serial-test.c static void check_guest_output(const testdef_t *test, int fd) · line 107

Quoted 8 lines
tests/boot-serial-test.c +1 -1
@@ -101,7 +101,7 @@ static void check_guest_output(const testdef_t *test, int fd)
     /* Poll serial output... Wait at most 60 seconds */
     for (i = 0; i < 6000; ++i) {
         ccnt = 0;
-        while ((nbr = read(fd, &ch, 1)) == 1 && ccnt++ < 512) {
+        while (ccnt++ < 512 && (nbr = read(fd, &ch, 1)) == 1) {
             if (ch == test->expect[pos]) {
                 pos += 1;
                 if (test->expect[pos] == '\0') {

Queued in the meanwhile, thanks.

Thanks, but Peter already applied it to fix the "make check" failures:

https://git.qemu.org/?p=qemu.git;a=commit;h=5e5432b766c424a5d1

Thomas