In task_init() there is a calculation of machdep->hz based on the value of
cfq_slice_async if the symbol exists:
if (symbol_exists("cfq_slice_async")) {
uint cfq_slice_async;
get_symbol_data("cfq_slice_async", sizeof(int),
&cfq_slice_async);
if (cfq_slice_async) {
machdep->hz = cfq_slice_async * 25;
if (CRASHDEBUG(2))
fprintf(fp,
"cfq_slice_async exists: setting hz to
%d\n",
machdep->hz);
}
}
However, patch "cfq-iosched: Convert from jiffies to nanoseconds"
(https://patchwork.kernel.org/project/linux-block/patch/[email protected]/)
changed the definition of cfq_slice_async from (HZ / 25) to (NSEC_PER_SEC /
25). As such, the above calculation will result in a value of 1000000000 for
machdep->hz. In vmcores where the symbol exists and has this definition, this
causes incorrect results for some calculations. I have a couple of 4.12
vmcores in this situation, and in them crash shows the uptime as 3 seconds,
which also throws off the timestamps in "log -T".
Fix this by skipping the above code block for kernels 4.8 and above:
--- task.c.orig 2021-04-14 16:07:52.021137849 -0400
+++ task.c 2021-04-14 16:11:03.406626370 -0400
@@ -417,7 +417,8 @@
STRUCT_SIZE_INIT(cputime_t, "cputime_t");
- if (symbol_exists("cfq_slice_async")) {
+ if ((THIS_KERNEL_VERSION < LINUX(4,8,0)) &&
+ symbol_exists("cfq_slice_async")) {
uint cfq_slice_async;
get_symbol_data("cfq_slice_async", sizeof(int),
Martin Moore
Linux Engineering Resolution Team
HPE Pointnext Services
Hewlett Packard Enterprise
--
Crash-utility mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/crash-utility
