In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
s390\/cpum_cf: Handle CPU hotplug via prepare\/dead callbacks<\/p>\n
The command ‘perf stat -e cycles — ‘ crashes the kernel
\nwhen CPUs are hotplug added during that run.<\/p>\n
Root cause is the allocation of struct cpu_cf_events at first
\nevent initialization. The allocation is dynamic and the first
\nevent that has task context creates such a structure for
\neach online CPU. This is not sufficient. CPUs may be offline
\nduring event creation and can be set online during the
\nperf run time. For example commands<\/p>\n
# echo 0 > \/sys\/devices\/system\/cpu\/cpu1\/online
\n # perf stat -e cycles -i — stress-ng -t10s –matrix X
\n # sleep 1
\n # echo 1 > \/sys\/devices\/system\/cpu\/cpu1\/online<\/p>\n
create an event for CPUs 0,2-X. Since the events are created with
\ntask-context, the scheduler will eventually schedule the program
\non CPU1. This CPU has not created and initialized any per
\nCPU event infrastructure as that CPU was not online at the time
\nof the perf invocation. Thus when the scheduler runs stress-ng
\non CPU1, the function cpumf_pmu_add() refers to a NULL pointer:<\/p>\n
struct cpu_cf_events *cpuhw = this_cpu_cfhw();<\/p>\n
This function call is invoked after the task stress-ng has been
\nmade runnable on CPU1. And this_cpu_cfhw() returns NULL.<\/p>\n
The result is a panic:
\nUnable to handle kernel pointer dereference in virtual kernel address space
\nFailing address: 0000000000000000 TEID: 0000000000000483
\n….
\nKrnl PSW : 0404d00180000000 000003ef8291fd0c (cpumf_pmu_add+0x3c\/0x80)
\n….
\nCall Trace:
\n [] cpumf_pmu_add+0x3c\/0x80
\n [] event_sched_in+0xae\/0x190
\n [] merge_sched_in+0x1b6\/0x390
\n [] visit_groups_merge.constprop.0.isra.0+0x308\/0x5b0
\n [] pmu_groups_sched_in+0x3a\/0x50
\n [] ctx_sched_in+0x180\/0x260
\n [] perf_event_context_sched_in+0x11c\/0x2d0
\n [] __perf_event_task_sched_in+0x2e\/0xc0
\n [] finish_task_switch.isra.0+0x1a4\/0x250
\n….
\nLast Breaking-Event-Address:
\n [] this_cpu_cfhw+0x38\/0x40<\/p>\n
The issue arises only in per-task context when the CPUMF facility is
\nused and the scheduler picks a random CPU for such a process to run on.
\nThe scheduler enables the CPUMF infrastructure via PMU callback
\nfunctions pmu::add() and pmu::del().<\/p>\n
Introduce a CPU hotplug prepare\/dead callback pair which creates and
\nremoves the per CPU counter data while the CPU is offline. Count the
\nusers which track every CPU (cpu == -1), that is perf_event_open()
\nevents with task context and \/dev\/hwctr device sessions, in the new
\ncounter cpu_cf_root::tskcnt, protected by pmc_reserve_mutex.
\nThis ensures the infrastructure is available when
\nnew CPU is selected to run the per-task context process.<\/p>\n
In cpum_cf_free_root() and cpum_cf_free_cpu() ensure the reference
\npointer to data structures is set to NULL before the data is freed
\nto prevent interrupt handlers to access stale data.<\/p>\n
[gor@linux.ibm.com: change commit message]<\/p>\n","protected":false},"excerpt":{"rendered":"
In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
s390\/cpum_cf: Handle CPU hotplug via prepare\/dead callbacks<\/p>\n
The command ‘perf stat -e cycles — ‘ crashes the kernel
\nwhen CPUs are<\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-21382","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/21382","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln"}],"about":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/types\/vuln"}],"wp:attachment":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=21382"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=21382"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=21382"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}