Public bug reported: We see the below failure in the LTP madvise07 test case, which tests HWPOISON functionality on anonymous mapped memory. The issue seems to be HWPOISON'ed anonymous mapped memory causes a process to be forcibly SIGKILL'd instead of being sent SIGBUS as expected.
~john-cabaj discovered this error while reviewing linux-azure-nvidia test results, and identified this patch as the culprit: https://git.launchpad.net/~canonical-kernel/ubuntu/+source/linux- nvidia/+git/noble/commit/?h=nvidia-6.14-next&id=7a1391d02f3e1a71b8953bfcd6a66c6093fe1280. This appears to only impact kernels based on noble:linux-nvidia-6.14. 77544 20:44:08 ERROR| [stderr] ------------------------------------------- 77545 20:44:08 ERROR| [stderr] INFO: runltp script is deprecated, try kirk 77546 20:44:08 ERROR| [stderr] https://github.com/linux-test-project/kirk 77547 20:44:08 ERROR| [stderr] ------------------------------------------- 77548 20:44:08 DEBUG| [stdout] Checking for required user/group ids 77549 20:44:08 DEBUG| [stdout] 77550 20:44:08 DEBUG| [stdout] 'root' user id and group found. 77551 20:44:08 DEBUG| [stdout] 'nobody' user id and group found. 77552 20:44:08 DEBUG| [stdout] 'bin' user id and group found. 77553 20:44:08 DEBUG| [stdout] 'daemon' user id and group found. 77554 20:44:08 DEBUG| [stdout] Users group found. 77555 20:44:08 DEBUG| [stdout] Sys group found. 77556 20:44:08 DEBUG| [stdout] Required users/groups exist. 77557 20:44:08 DEBUG| [stdout] no big block device was specified on commandline. 77558 20:44:08 DEBUG| [stdout] Tests which require a big block device are disabled. 77559 20:44:08 DEBUG| [stdout] You can specify it with option -z 77560 20:44:08 DEBUG| [stdout] INFO: Test start time: Thu Sep 18 20:44:08 UTC 2025 77561 20:44:08 DEBUG| [stdout] COMMAND: /opt/ltp/bin/ltp-pan -q -e -S -a 150147 -n 150147 -f /tmp/ltp-1RFGjDBAMw/alltests -l /dev/null -C /dev/null -T /dev/null 77562 20:44:08 DEBUG| [stdout] LOG File: /dev/null 77563 20:44:08 DEBUG| [stdout] FAILED COMMAND File: /dev/null 77564 20:44:08 DEBUG| [stdout] TCONF COMMAND File: /dev/null 77565 20:44:08 DEBUG| [stdout] Running tests....... 77566 20:44:08 DEBUG| [stdout] tst_test.c:1952: TINFO: LTP version: 20250130 77567 20:44:08 DEBUG| [stdout] tst_test.c:1955: TINFO: Tested kernel: 6.14.0-1004-azure-nvidia #4~24.04.1-Ubuntu SMP Thu Sep 11 18:20:32 UTC 2025 aarch64 77568 20:44:08 DEBUG| [stdout] tst_kconfig.c:88: TINFO: Parsing kernel config '/lib/modules/6.14.0-1004-azure-nvidia/build/.config' 77569 20:44:08 DEBUG| [stdout] tst_test.c:1773: TINFO: Overall timeout per run is 0h 00m 30s 77570 20:44:08 DEBUG| [stdout] madvise07.c:43: TINFO: mmap(0, 65536, PROT_READ | PROT_WRITE, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0) 77571 20:44:08 DEBUG| [stdout] madvise07.c:54: TINFO: madvise(0xe595bee90000, 65536, MADV_HWPOISON) 77572 20:44:08 DEBUG| [stdout] madvise07.c:90: TFAIL: Child killed by SIGKILL 77573 20:44:08 DEBUG| [stdout] 77574 20:44:08 DEBUG| [stdout] Summary: 77575 20:44:08 DEBUG| [stdout] passed 0 77576 20:44:08 DEBUG| [stdout] failed 1 77577 20:44:08 DEBUG| [stdout] broken 0 77578 20:44:08 DEBUG| [stdout] skipped 0 77579 20:44:08 DEBUG| [stdout] warnings 0 77580 20:44:08 DEBUG| [stdout] INFO: ltp-pan reported some tests FAIL 77581 20:44:08 DEBUG| [stdout] LTP Version: 20250130 77582 20:44:08 DEBUG| [stdout] INFO: Test end time: Thu Sep 18 20:44:08 UTC 2025 77583 20:44:08 ERROR| [stderr] ------------------------------------------- 77584 20:44:08 ERROR| [stderr] INFO: runltp script is deprecated, try kirk 77585 20:44:08 ERROR| [stderr] https://github.com/linux-test-project/kirk 77586 20:44:08 ERROR| [stderr] ------------------------------------------- This is accompanied by an error in dmesg: ... [ 2786.499480] Memory failure: 0x232b9: forcibly killing madvise07:150194 because of failure to unmap corrupted page ... ** Affects: linux-nvidia-6.14 (Ubuntu) Importance: Undecided Status: Invalid ** Affects: linux-nvidia-6.14 (Ubuntu Noble) Importance: Undecided Status: Triaged ** Also affects: linux-nvidia-6.14 (Ubuntu Noble) Importance: Undecided Status: New -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux-nvidia-6.14 in Ubuntu. https://bugs.launchpad.net/bugs/2125434 Title: 6.14 kernel SAUCE patch causes LTP madvise07 test case to fail with "Child killed by SIGKILL" Status in linux-nvidia-6.14 package in Ubuntu: Invalid Status in linux-nvidia-6.14 source package in Noble: Triaged Bug description: We see the below failure in the LTP madvise07 test case, which tests HWPOISON functionality on anonymous mapped memory. The issue seems to be HWPOISON'ed anonymous mapped memory causes a process to be forcibly SIGKILL'd instead of being sent SIGBUS as expected. ~john-cabaj discovered this error while reviewing linux-azure-nvidia test results, and identified this patch as the culprit: https://git.launchpad.net/~canonical-kernel/ubuntu/+source/linux- nvidia/+git/noble/commit/?h=nvidia-6.14-next&id=7a1391d02f3e1a71b8953bfcd6a66c6093fe1280. This appears to only impact kernels based on noble:linux-nvidia-6.14. 77544 20:44:08 ERROR| [stderr] ------------------------------------------- 77545 20:44:08 ERROR| [stderr] INFO: runltp script is deprecated, try kirk 77546 20:44:08 ERROR| [stderr] https://github.com/linux-test-project/kirk 77547 20:44:08 ERROR| [stderr] ------------------------------------------- 77548 20:44:08 DEBUG| [stdout] Checking for required user/group ids 77549 20:44:08 DEBUG| [stdout] 77550 20:44:08 DEBUG| [stdout] 'root' user id and group found. 77551 20:44:08 DEBUG| [stdout] 'nobody' user id and group found. 77552 20:44:08 DEBUG| [stdout] 'bin' user id and group found. 77553 20:44:08 DEBUG| [stdout] 'daemon' user id and group found. 77554 20:44:08 DEBUG| [stdout] Users group found. 77555 20:44:08 DEBUG| [stdout] Sys group found. 77556 20:44:08 DEBUG| [stdout] Required users/groups exist. 77557 20:44:08 DEBUG| [stdout] no big block device was specified on commandline. 77558 20:44:08 DEBUG| [stdout] Tests which require a big block device are disabled. 77559 20:44:08 DEBUG| [stdout] You can specify it with option -z 77560 20:44:08 DEBUG| [stdout] INFO: Test start time: Thu Sep 18 20:44:08 UTC 2025 77561 20:44:08 DEBUG| [stdout] COMMAND: /opt/ltp/bin/ltp-pan -q -e -S -a 150147 -n 150147 -f /tmp/ltp-1RFGjDBAMw/alltests -l /dev/null -C /dev/null -T /dev/null 77562 20:44:08 DEBUG| [stdout] LOG File: /dev/null 77563 20:44:08 DEBUG| [stdout] FAILED COMMAND File: /dev/null 77564 20:44:08 DEBUG| [stdout] TCONF COMMAND File: /dev/null 77565 20:44:08 DEBUG| [stdout] Running tests....... 77566 20:44:08 DEBUG| [stdout] tst_test.c:1952: TINFO: LTP version: 20250130 77567 20:44:08 DEBUG| [stdout] tst_test.c:1955: TINFO: Tested kernel: 6.14.0-1004-azure-nvidia #4~24.04.1-Ubuntu SMP Thu Sep 11 18:20:32 UTC 2025 aarch64 77568 20:44:08 DEBUG| [stdout] tst_kconfig.c:88: TINFO: Parsing kernel config '/lib/modules/6.14.0-1004-azure-nvidia/build/.config' 77569 20:44:08 DEBUG| [stdout] tst_test.c:1773: TINFO: Overall timeout per run is 0h 00m 30s 77570 20:44:08 DEBUG| [stdout] madvise07.c:43: TINFO: mmap(0, 65536, PROT_READ | PROT_WRITE, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0) 77571 20:44:08 DEBUG| [stdout] madvise07.c:54: TINFO: madvise(0xe595bee90000, 65536, MADV_HWPOISON) 77572 20:44:08 DEBUG| [stdout] madvise07.c:90: TFAIL: Child killed by SIGKILL 77573 20:44:08 DEBUG| [stdout] 77574 20:44:08 DEBUG| [stdout] Summary: 77575 20:44:08 DEBUG| [stdout] passed 0 77576 20:44:08 DEBUG| [stdout] failed 1 77577 20:44:08 DEBUG| [stdout] broken 0 77578 20:44:08 DEBUG| [stdout] skipped 0 77579 20:44:08 DEBUG| [stdout] warnings 0 77580 20:44:08 DEBUG| [stdout] INFO: ltp-pan reported some tests FAIL 77581 20:44:08 DEBUG| [stdout] LTP Version: 20250130 77582 20:44:08 DEBUG| [stdout] INFO: Test end time: Thu Sep 18 20:44:08 UTC 2025 77583 20:44:08 ERROR| [stderr] ------------------------------------------- 77584 20:44:08 ERROR| [stderr] INFO: runltp script is deprecated, try kirk 77585 20:44:08 ERROR| [stderr] https://github.com/linux-test-project/kirk 77586 20:44:08 ERROR| [stderr] ------------------------------------------- This is accompanied by an error in dmesg: ... [ 2786.499480] Memory failure: 0x232b9: forcibly killing madvise07:150194 because of failure to unmap corrupted page ... To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux-nvidia-6.14/+bug/2125434/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : [email protected] Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
