ktest: add 2nd parameter of run_command() to set the redirect target file

From: Satoru Takeuchi <[email protected]>

Build failure hapens with the following log if BUILD_DIR is dirty.

===============================================================================
...
RUNNING TEST 1 of 1 with option build olddefconfig

make O=/home/sat/ktest/output mrproper ... CLEAN arch/x86/tools

  CLEAN   scripts/basic
  CLEAN   scripts/kconfig
  CLEAN   include/config include/generated arch/x86/include/generated
  CLEAN   .config.old include/generated/uapi/linux/version.h

SUCCESS
...
make O=/home/sat/ktest/output -j8 ... GEN /home/sat/ktest/output/Makefile
...
Using /home/sat/src/test-linux-stable/linux-stable as source for kernel /home/sat/src/test-linux-stable/linux-stable is not clean, please run 'make mrproper' in the '/home/sat/src/test-linux-stable/linux-stable' directory.
make[1]: *** [prepare3] Error 1
make[1]: *** Waiting for unfinished jobs....
HOSTLD arch/x86/tools/relocs
make[1]: *** wait: No child processes. Stop.
make: *** [sub-make] Error 2
FAILED!
CRITICAL FAILURE... failed build
===============================================================================

To avoid this problem, it's necessary to run make mrproper not only for OUTPUT_DIR, but also for BUILD_DIR.

Signed-off-by Satoru Takeuchi <[email protected]>

Cc Steven Rostedt <[email protected]>

---
 tools/testing/ktest/ktest.pl | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

tools/testing/ktest/ktest.pl +9 -7

index f731ef6..1169b2a 100755
@@ -120,6 +120,7 @@ my $syslinux;
 my $syslinux_path;
 my $syslinux_label;
 my $target;
+my $makecmd;
 my $make;
 my $pre_install;
 my $post_install;

@@ -2198,6 +2199,11 @@ sub load_force_config {
     close IN;
 }
 
+sub mrproper {
+    run_command "$makecmd mrproper" or dodie "make mrproper for $builddir";
+    run_command "$make mrproper" or dodie "make mrproper for $outputdir";
+}
+
 sub build {
     my ($type) = @_;
 

@@ -2235,17 +2241,13 @@ sub build {
 	if (!$noclean) {
 	    run_command "mv $output_config $outputdir/config_temp" or
 		dodie "moving .config";
-
-	    run_command "$make mrproper" or dodie "make mrproper";
-
+	    mrproper;
 	    run_command "mv $outputdir/config_temp $output_config" or
 		dodie "moving config_temp";
 	}
-
     } elsif (!$noclean) {
 	unlink "$output_config";
-	run_command "$make mrproper" or
-	    dodie "make mrproper";
+	mrproper;
     }
 
     # add something to distinguish this build

@@ -3996,7 +3998,7 @@ for (my $i = 1; $i <= $opt{"NUM_TESTS"}; $i++) {
 
     undef %force_config;
 
-    my $makecmd = set_test_option("MAKE_CMD", $i);
+    $makecmd = set_test_option("MAKE_CMD", $i);
 
     $outputdir = set_test_option("OUTPUT_DIR", $i);
     $builddir = set_test_option("BUILD_DIR", $i);

Quoted 3 lines · On Sun, 02 Mar 2014 21:23:36 +0900 Satoru Takeuchi <[email protected]>…

From: Satoru Takeuchi <[email protected]>

Build failure hapens with the following log if BUILD_DIR is dirty.

If you want to do make mrproper in the BUILD_DIR, add the command to PRE_BUILD or something. I have TAGS in my BUILD_DIRs and this will remove them for me. I can't accept this change.

I've noticed that there's a bug in the Linux make system that causes the O=.. builds to fail because somehow the source directory got dirty. We need to figure out what that bug is.

Thanks,

-- Steve

Quoted 89 lines

===============================================================================
...
RUNNING TEST 1 of 1 with option build olddefconfig

make O=/home/sat/ktest/output mrproper ... CLEAN arch/x86/tools

  CLEAN   scripts/basic
  CLEAN   scripts/kconfig
  CLEAN   include/config include/generated arch/x86/include/generated
  CLEAN   .config.old include/generated/uapi/linux/version.h

SUCCESS
...
make O=/home/sat/ktest/output -j8 ... GEN /home/sat/ktest/output/Makefile
...
Using /home/sat/src/test-linux-stable/linux-stable as source for kernel /home/sat/src/test-linux-stable/linux-stable is not clean, please run 'make mrproper' in the '/home/sat/src/test-linux-stable/linux-stable' directory.
make[1]: *** [prepare3] Error 1
make[1]: *** Waiting for unfinished jobs....
HOSTLD arch/x86/tools/relocs
make[1]: *** wait: No child processes. Stop.
make: *** [sub-make] Error 2
FAILED!
CRITICAL FAILURE... failed build
===============================================================================

To avoid this problem, it's necessary to run make mrproper not only for OUTPUT_DIR, but also for BUILD_DIR.

Signed-off-by Satoru Takeuchi <[email protected]>

Cc Steven Rostedt <[email protected]>

---
 tools/testing/ktest/ktest.pl | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

tools/testing/ktest/ktest.pl +9 -7

index f731ef6..1169b2a 100755
@@ -120,6 +120,7 @@ my $syslinux;
 my $syslinux_path;
 my $syslinux_label;
 my $target;
+my $makecmd;
 my $make;
 my $pre_install;
 my $post_install;

@@ -2198,6 +2199,11 @@ sub load_force_config {
     close IN;
 }
 
+sub mrproper {
+    run_command "$makecmd mrproper" or dodie "make mrproper for $builddir";
+    run_command "$make mrproper" or dodie "make mrproper for $outputdir";
+}
+
 sub build {
     my ($type) = @_;
 

@@ -2235,17 +2241,13 @@ sub build {
 	if (!$noclean) {
 	    run_command "mv $output_config $outputdir/config_temp" or
 		dodie "moving .config";
-
-	    run_command "$make mrproper" or dodie "make mrproper";
-
+	    mrproper;
 	    run_command "mv $outputdir/config_temp $output_config" or
 		dodie "moving config_temp";
 	}
-
     } elsif (!$noclean) {
 	unlink "$output_config";
-	run_command "$make mrproper" or
-	    dodie "make mrproper";
+	mrproper;
     }
 
     # add something to distinguish this build

@@ -3996,7 +3998,7 @@ for (my $i = 1; $i <= $opt{"NUM_TESTS"}; $i++) {
 
     undef %force_config;
 
-    my $makecmd = set_test_option("MAKE_CMD", $i);
+    $makecmd = set_test_option("MAKE_CMD", $i);
 
     $outputdir = set_test_option("OUTPUT_DIR", $i);
     $builddir = set_test_option("BUILD_DIR", $i);

At Sun, 2 Mar 2014 08:29:16 -0500,

Quoted 12 lines · Steven Rostedt

Quoted 3 lines · On Sun, 02 Mar 2014 21:23:36 +0900 Satoru Takeuchi <[email protected]>…

From: Satoru Takeuchi <[email protected]>

Build failure hapens with the following log if BUILD_DIR is dirty.

If you want to do make mrproper in the BUILD_DIR, add the command to PRE_BUILD or something. I have TAGS in my BUILD_DIRs and this will remove them for me. I can't accept this change.

Quoted 3 lines

I've noticed that there's a bug in the Linux make system that causes the O=.. builds to fail because somehow the source directory got dirty. We need to figure out what that bug is.

I see. So what I should fix is kbuild system.

From: Satoru Takeuchi <[email protected]>

If we'd like to set the redirect target file of run_command(),
we should define $redirect before calling this function and should undef it after calling this function. Since it's user-unfriendly, add 2nd parameter of run_command() for this purpose.

Signed-off-by Satoru Takeuchi <[email protected]>

Cc Steven Rostedt <[email protected]>

[PATCH 2/2] ktest: run make mrproper not only for OUTPUT_DIR but also for BUILD_DIR to avoid build failure

---
 tools/testing/ktest/ktest.pl | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)

tools/testing/ktest/ktest.pl +3 -8

index 4063156..f731ef6 100755
@@ -149,7 +149,6 @@ my $bisect_ret_abort;
 my $bisect_ret_default;
 my $in_patchcheck = 0;
 my $run_test;
-my $redirect;
 my $buildlog;
 my $testlog;
 my $dmesg;

@@ -1529,7 +1528,7 @@ sub fail {
 }
 
 sub run_command {
-    my ($command) = @_;
+    my ($command, $redirect) = @_;
     my $dolog = 0;
     my $dord = 0;
     my $pid;

@@ -2265,9 +2264,7 @@ sub build {
     # Run old config regardless, to enforce min configurations
     make_oldconfig;
 
-    $redirect = "$buildlog";
-    my $build_ret = run_command "$make $build_options";
-    undef $redirect;
+    my $build_ret = run_command "$make $build_options", $buildlog;
 
     if (defined($post_build)) {
 	# Because a post build may change the kernel version

@@ -2360,9 +2357,7 @@ sub child_run_test {
     $poweroff_on_error = 0;
     $die_on_failure = 1;
 
-    $redirect = "$testlog";
-    run_command $run_test or $failed = 1;
-    undef $redirect;
+    run_command $run_test, $testlog or $failed = 1;
 
     exit $failed;
 }

Signature · 1 line

1.8.5.3

Hi Steven,

At Sun, 02 Mar 2014 21:20:31 +0900,

Quoted 8 lines · Satoru Takeuchi

From: Satoru Takeuchi <[email protected]>

If we'd like to set the redirect target file of run_command(),
we should define $redirect before calling this function and should undef it after calling this function. Since it's user-unfriendly, add 2nd parameter of run_command() for this purpose.

Quoted 2 lines

Signed-off-by Satoru Takeuchi <[email protected]>

Cc Steven Rostedt <[email protected]>

Could you tell whether this patch will be applied? Although you told me that you won't apply the following patch, I didn't hear about this patch.

Quoted 2 lines

[PATCH 2/2] ktest: run make mrproper not only for OUTPUT_DIR but also for BUILD_DIR to avoid build failure

I make this question since I'll send next patch set this weekend and want to know whether I can make patch on the top of this patch.

Quoted 12 lines · On Sat, 08 Mar 2014 09:12:31 +0900 Satoru Takeuchi <[email protected]>…

Hi Steven,

At Sun, 02 Mar 2014 21:20:31 +0900,

Quoted 8 lines · Satoru Takeuchi

From: Satoru Takeuchi <[email protected]>

If we'd like to set the redirect target file of run_command(),
we should define $redirect before calling this function and should undef it after calling this function. Since it's user-unfriendly, add 2nd parameter of run_command() for this purpose.

Quoted 5 lines

Quoted 2 lines

Signed-off-by Satoru Takeuchi <[email protected]>

Cc Steven Rostedt <[email protected]>

Could you tell whether this patch will be applied? Although you told me that you won't apply the following patch, I didn't hear about this patch.

Ah, I was thinking this one was needed for the next patch, so I didn't apply it.

Quoted 6 lines

Quoted 3 lines

[PATCH 2/2] ktest: run make mrproper not only for OUTPUT_DIR but also for BUILD_DIR to avoid build failure

I make this question since I'll send next patch set this weekend and want to know whether I can make patch on the top of this patch.

You can add this one again. I haven't really reviewed it, but if other patches depend on it, include it as well. Include it anyway if you want me to review it.

Thanks,

-- Steve