| Message ID | 20220404182137.59231-1-hbathini@linux.ibm.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
| Series | powerpc/fadump: save CPU reg data in vmcore when PHYP terminates LPAR | expand |
| Context | Check | Description |
|---|---|---|
| snowpatch_ozlabs/github-powerpc_ppctests | success | Successfully ran 8 jobs. |
| snowpatch_ozlabs/github-powerpc_selftests | success | Successfully ran 8 jobs. |
| snowpatch_ozlabs/github-powerpc_kernel_qemu | success | Successfully ran 24 jobs. |
| snowpatch_ozlabs/github-powerpc_sparse | success | Successfully ran 4 jobs. |
| snowpatch_ozlabs/github-powerpc_clang | success | Successfully ran 7 jobs. |
On 2022-04-04 23:51:37 Mon, Hari Bathini wrote: > An LPAR can be terminated by the POWER Hypervisor (PHYP) for various > reasons. If FADump was configured when PHYP terminates the LPAR, > platform-assisted dump is initiated to save the kernel dump. But CPU > register data would not be processed/saved in the vmcore in such case > because CPU mask is set in crash_fadump() at the time of kernel crash > and it remains unset in this case with LPAR being terminated by PHYP > abruptly. > > To get around the problem, initialize cpu_mask to cpu_possible_mask > so as to ensure all possible CPUs' register data is processed for the > vmcore generated on PHYP terminated LPAR. Also, rename the crash info > member variable from online_mask to cpu_mask as it doesn't necessarily > have to be online CPU mask always. > > Signed-off-by: Hari Bathini <hbathini@linux.ibm.com> > --- > arch/powerpc/include/asm/fadump-internal.h | 2 +- > arch/powerpc/kernel/fadump.c | 7 ++++++- > arch/powerpc/platforms/pseries/rtas-fadump.c | 2 +- > 3 files changed, 8 insertions(+), 3 deletions(-) > > diff --git a/arch/powerpc/include/asm/fadump-internal.h b/arch/powerpc/include/asm/fadump-internal.h > index 81bcb9abb371..27f9e11eda28 100644 > --- a/arch/powerpc/include/asm/fadump-internal.h > +++ b/arch/powerpc/include/asm/fadump-internal.h > @@ -50,7 +50,7 @@ struct fadump_crash_info_header { > u64 elfcorehdr_addr; > u32 crashing_cpu; > struct pt_regs regs; > - struct cpumask online_mask; > + struct cpumask cpu_mask; > }; > > struct fadump_memory_range { > diff --git a/arch/powerpc/kernel/fadump.c b/arch/powerpc/kernel/fadump.c > index 65562c4a0a69..8343c0b14277 100644 > --- a/arch/powerpc/kernel/fadump.c > +++ b/arch/powerpc/kernel/fadump.c > @@ -728,7 +728,7 @@ void crash_fadump(struct pt_regs *regs, const char *str) > else > ppc_save_regs(&fdh->regs); > > - fdh->online_mask = *cpu_online_mask; > + fdh->cpu_mask = *cpu_online_mask; > > /* > * If we came in via system reset, wait a while for the secondary > @@ -1164,6 +1164,11 @@ static unsigned long init_fadump_header(unsigned long addr) > fdh->elfcorehdr_addr = addr; > /* We will set the crashing cpu id in crash_fadump() during crash. */ > fdh->crashing_cpu = FADUMP_CPU_UNKNOWN; > + /* > + * When LPAR is terminated by PYHP, ensure all possible CPUs' > + * register data is processed while exporting the vmcore. > + */ > + fdh->cpu_mask = *cpu_possible_mask; Looks good to me. Reviewed-by: Mahesh Salgaonkar <mahesh@linux.ibm.com> Thanks, -Mahesh. > > return addr; > } > diff --git a/arch/powerpc/platforms/pseries/rtas-fadump.c b/arch/powerpc/platforms/pseries/rtas-fadump.c > index 35f9cb602c30..617c0f3b1f4f 100644 > --- a/arch/powerpc/platforms/pseries/rtas-fadump.c > +++ b/arch/powerpc/platforms/pseries/rtas-fadump.c > @@ -351,7 +351,7 @@ static int __init rtas_fadump_build_cpu_notes(struct fw_dump *fadump_conf) > /* Lower 4 bytes of reg_value contains logical cpu id */ > cpu = (be64_to_cpu(reg_entry->reg_value) & > RTAS_FADUMP_CPU_ID_MASK); > - if (fdh && !cpumask_test_cpu(cpu, &fdh->online_mask)) { > + if (fdh && !cpumask_test_cpu(cpu, &fdh->cpu_mask)) { > RTAS_FADUMP_SKIP_TO_NEXT_CPU(reg_entry); > continue; > } > -- > 2.35.1 >
On Mon, 4 Apr 2022 23:51:37 +0530, Hari Bathini wrote: > An LPAR can be terminated by the POWER Hypervisor (PHYP) for various > reasons. If FADump was configured when PHYP terminates the LPAR, > platform-assisted dump is initiated to save the kernel dump. But CPU > register data would not be processed/saved in the vmcore in such case > because CPU mask is set in crash_fadump() at the time of kernel crash > and it remains unset in this case with LPAR being terminated by PHYP > abruptly. > > [...] Applied to powerpc/next. [1/1] powerpc/fadump: save CPU reg data in vmcore when PHYP terminates LPAR https://git.kernel.org/powerpc/c/6584cec0a2255ab407d047d1b135fa0aae88d6c6 cheers
diff --git a/arch/powerpc/include/asm/fadump-internal.h b/arch/powerpc/include/asm/fadump-internal.h index 81bcb9abb371..27f9e11eda28 100644 --- a/arch/powerpc/include/asm/fadump-internal.h +++ b/arch/powerpc/include/asm/fadump-internal.h @@ -50,7 +50,7 @@ struct fadump_crash_info_header { u64 elfcorehdr_addr; u32 crashing_cpu; struct pt_regs regs; - struct cpumask online_mask; + struct cpumask cpu_mask; }; struct fadump_memory_range { diff --git a/arch/powerpc/kernel/fadump.c b/arch/powerpc/kernel/fadump.c index 65562c4a0a69..8343c0b14277 100644 --- a/arch/powerpc/kernel/fadump.c +++ b/arch/powerpc/kernel/fadump.c @@ -728,7 +728,7 @@ void crash_fadump(struct pt_regs *regs, const char *str) else ppc_save_regs(&fdh->regs); - fdh->online_mask = *cpu_online_mask; + fdh->cpu_mask = *cpu_online_mask; /* * If we came in via system reset, wait a while for the secondary @@ -1164,6 +1164,11 @@ static unsigned long init_fadump_header(unsigned long addr) fdh->elfcorehdr_addr = addr; /* We will set the crashing cpu id in crash_fadump() during crash. */ fdh->crashing_cpu = FADUMP_CPU_UNKNOWN; + /* + * When LPAR is terminated by PYHP, ensure all possible CPUs' + * register data is processed while exporting the vmcore. + */ + fdh->cpu_mask = *cpu_possible_mask; return addr; } diff --git a/arch/powerpc/platforms/pseries/rtas-fadump.c b/arch/powerpc/platforms/pseries/rtas-fadump.c index 35f9cb602c30..617c0f3b1f4f 100644 --- a/arch/powerpc/platforms/pseries/rtas-fadump.c +++ b/arch/powerpc/platforms/pseries/rtas-fadump.c @@ -351,7 +351,7 @@ static int __init rtas_fadump_build_cpu_notes(struct fw_dump *fadump_conf) /* Lower 4 bytes of reg_value contains logical cpu id */ cpu = (be64_to_cpu(reg_entry->reg_value) & RTAS_FADUMP_CPU_ID_MASK); - if (fdh && !cpumask_test_cpu(cpu, &fdh->online_mask)) { + if (fdh && !cpumask_test_cpu(cpu, &fdh->cpu_mask)) { RTAS_FADUMP_SKIP_TO_NEXT_CPU(reg_entry); continue; }
An LPAR can be terminated by the POWER Hypervisor (PHYP) for various reasons. If FADump was configured when PHYP terminates the LPAR, platform-assisted dump is initiated to save the kernel dump. But CPU register data would not be processed/saved in the vmcore in such case because CPU mask is set in crash_fadump() at the time of kernel crash and it remains unset in this case with LPAR being terminated by PHYP abruptly. To get around the problem, initialize cpu_mask to cpu_possible_mask so as to ensure all possible CPUs' register data is processed for the vmcore generated on PHYP terminated LPAR. Also, rename the crash info member variable from online_mask to cpu_mask as it doesn't necessarily have to be online CPU mask always. Signed-off-by: Hari Bathini <hbathini@linux.ibm.com> --- arch/powerpc/include/asm/fadump-internal.h | 2 +- arch/powerpc/kernel/fadump.c | 7 ++++++- arch/powerpc/platforms/pseries/rtas-fadump.c | 2 +- 3 files changed, 8 insertions(+), 3 deletions(-)