Fix kdump/Fadump config

ServiceReport fails to estimate correct crashkernel size. This is a temporary fix to define crashkernel size. Signed-off-by: Sachin P Bappalige <sachinpb@linux.ibm.com>
open-power · Apr 24, 2024 · 1861545 · 1861545
1 parent e4c4fb6
commit 1861545
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 11 deletions.
diff --git a/op-test b/op-test
@@ -693,15 +693,15 @@ class OSdumpfadumpSuite():
     def __init__(self):
         self.s = unittest.TestSuite()
         self.s.addTest(PowerNVDump.KernelCrash_FadumpEnable())
-        self.s.addTest(PowerNVDump.OpTestMakedump())
         self.s.addTest(PowerNVDump.KernelCrash_KdumpDLPAR())
         self.s.addTest(PowerNVDump.KernelCrash_KdumpWorkLoad())
         self.s.addTest(PowerNVDump.KernelCrash_hugepage_checks())
         self.s.addTest(PowerNVDump.KernelCrash_XIVE_off())
         self.s.addTest(PowerNVDump.KernelCrash_disable_radix())
-        self.s.addTest(PowerNVDump.KernelCrash_FadumpNocma())
+        self.s.addTest(PowerNVDump.OpTestMakedump())
         self.s.addTest(PowerNVDump.KernelCrash_KdumpSSH())
         self.s.addTest(PowerNVDump.KernelCrash_KdumpNFS())
+        self.s.addTest(PowerNVDump.KernelCrash_FadumpNocma())
 
     def suite(self):
         return self.s

diff --git a/testcases/PowerNVDump.py b/testcases/PowerNVDump.py
@@ -294,8 +294,7 @@ def verify_fadump_reg(self):
         if int(res) == 1:
             return True 
         else:
-            self.c.run_command("echo 1 > /sys/kernel/fadump_registered")
-            self.c.run_command('\r\n')
+            self.c.run_command("echo 1 > /sys/kernel/fadump_registered; sleep 10")
             self.c.run_command("cat /sys/kernel/fadump_registered")
 
         if not self.is_lpar:
@@ -330,14 +329,12 @@ def verify_fadump_unreg(self):
         if int(res) == 0:
             return True 
         else:
-            self.c.run_command("echo 0 > /sys/kernel/fadump_registered")
-            self.c.run_command('\r\n')
+            self.c.run_command("echo 0 > /sys/kernel/fadump_registered; sleep 10")
 
         if not self.is_lpar:
             self.c.run_command("%s > /tmp/opal_log" % BMC_CONST.OPAL_MSG_LOG)
             self.c.run_command("dmesg > /tmp/dmesg_log")
-            self.c.run_command("echo 0 > /sys/kernel/fadump_registered")
-            self.c.run_command('\r\n')
+            self.c.run_command("echo 0 > /sys/kernel/fadump_registered; sleep 10")
 
             opal_data = " ".join(self.c.run_command(
                 "%s | diff -a /tmp/opal_log -" % BMC_CONST.OPAL_MSG_LOG))
@@ -632,7 +629,7 @@ def setup_fadump(self):
         '''
         self.cv_SYSTEM.set_state(OpSystemState.OS)
         if self.distro == "rhel":
-            self.c.run_command("git clone https://github.com/linux-ras/ServiceReport; cd ServiceReport;"
+            self.c.run_command("rm -rf ServiceReport; git clone https://github.com/linux-ras/ServiceReport; cd ServiceReport;"
                                "python ./servicereport --plugins kdump package --repair", timeout=240)
             time.sleep(10)
             self.c.run_command("sed -e '/nfs/ s/^#*/#/' -i /etc/kdump.conf; sync")
@@ -718,7 +715,7 @@ def runTest(self):
             self.cv_HOST.host_check_command("kdump")
         elif self.distro == "rhel":
             self.cv_HOST.host_check_command("kdumpctl")
-            self.c.run_command("git clone https://github.com/linux-ras/ServiceReport; cd ServiceReport;"
+            self.c.run_command("rm -rf ServiceReport; git clone https://github.com/linux-ras/ServiceReport; cd ServiceReport;"
                                "python ./servicereport --plugins kdump package --repair", timeout=240)
             time.sleep(10)
         elif self.distro == "sles":
@@ -733,9 +730,11 @@ def runTest(self):
         self.cv_HOST.host_enable_kdump_service(os_level)
         log.info("========= Testing Only kdump enable followed by crash ===========")
         boot_type = self.kernel_crash()
+        self.verify_dump_file(boot_type)
         if self.is_lpar:
+            log.info("========= Testing kdump with HMC dumprestart ===========")
             boot_type = self.kernel_crash(crash_type="hmc")
-        self.verify_dump_file(boot_type)
+            self.verify_dump_file(boot_type)
 
 
 class KernelCrash_DisableAll(PowerNVDump):
@@ -1127,6 +1126,10 @@ class KernelCrash_XIVE_off(PowerNVDump):
     '''
 
     def runTest(self):
+        obj = OpTestInstallUtil.InstallUtil()
+        if not obj.update_kernel_cmdline(self.distro, remove_args="default_hugepagesz=1GB hugepagesz=1GB hugepages=80",
+                                         reboot=True, reboot_cmd=True):
+            self.fail("KernelArgTest failed to remove kernel args:default_hugepagesz=1GB hugepagesz=1GB hugepages=80")
         self.cv_SYSTEM.goto_state(OpSystemState.OS)
         self.setup_test()
         log.info("=============== Testing kdump/fadump with xive=off ===============")
@@ -1171,6 +1174,10 @@ class KernelCrash_disable_radix(PowerNVDump):
     '''
 
     def runTest(self):
+        obj = OpTestInstallUtil.InstallUtil()
+        if not obj.update_kernel_cmdline(self.distro, remove_args="xive=off",
+                                         reboot=True, reboot_cmd=True):
+           self.fail("KernelArgTest failed to remove kernel args:xive=off")
         self.cv_SYSTEM.goto_state(OpSystemState.OS)
         self.setup_test()
         log.info("Testing kdump/fadump with disable_radix")
@@ -1279,6 +1286,10 @@ def makedump_check(self):
         self.c.run_command("rm -rf dump*")
 
     def runTest(self):
+        obj = OpTestInstallUtil.InstallUtil()
+        if not obj.update_kernel_cmdline(self.distro, remove_args="disable_radix",
+                                         reboot=True, reboot_cmd=True):
+            self.fail("KernelArgTest failed to remove kernel args:disable_radix")
         self.kernel_crash()
         self.makedump_check()