diff mbox series

[v19,7/7] x86/crash: optimize cpu changes

Message ID 20230306162228.8277-8-eric.devolder@oracle.com
State New
Headers
Series crash: Kernel handling of CPU and memory hot un/plug |

Commit Message

Eric DeVolder March 6, 2023, 4:22 p.m. UTC 
  This patch is dependent upon the patch 'crash: change
crash_prepare_elf64_headers() to for_each_possible_cpu()'. With that
patch, crash_prepare_elf64_headers() writes out an ELF cpu PT_NOTE
for all possible cpus, thus further cpu changes to the elfcorehdr
are not needed.

This change works for kexec_file_load() and kexec_load() syscalls.
For kexec_file_load(), crash_prepare_elf64_headers() is utilized
directly and thus all ELF cpu PT_NOTEs are in the elfcorehdr already.
This is the kimage->file_mode term.
For kexec_load() syscall, one cpu or memory change will cause the
elfcorehdr to be updated via crash_prepare_elf64_headers() and at
that point all ELF cpu PT_NOTEs are in the elfcorehdr. This is the
kimage->elfcorehdr_updated term.

This code is intentionally *NOT* hoisted into
crash_handle_hotplug_event() as it would prevent the arch-specific
handler from running for cpu changes. This would break PPC, for
example, which needs to update other information besides the
elfcorehdr, on cpu changes.

Signed-off-by: Eric DeVolder <eric.devolder@oracle.com>
---
 arch/x86/kernel/crash.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

Comments

Sourabh Jain March 7, 2023, 9 a.m. UTC | #1 
On 06/03/23 21:52, Eric DeVolder wrote:
> This patch is dependent upon the patch 'crash: change
> crash_prepare_elf64_headers() to for_each_possible_cpu()'. With that
> patch, crash_prepare_elf64_headers() writes out an ELF cpu PT_NOTE
> for all possible cpus, thus further cpu changes to the elfcorehdr
> are not needed.
>
> This change works for kexec_file_load() and kexec_load() syscalls.
> For kexec_file_load(), crash_prepare_elf64_headers() is utilized
> directly and thus all ELF cpu PT_NOTEs are in the elfcorehdr already.
> This is the kimage->file_mode term.
> For kexec_load() syscall, one cpu or memory change will cause the
> elfcorehdr to be updated via crash_prepare_elf64_headers() and at
> that point all ELF cpu PT_NOTEs are in the elfcorehdr. This is the
> kimage->elfcorehdr_updated term.
>
> This code is intentionally *NOT* hoisted into
> crash_handle_hotplug_event() as it would prevent the arch-specific
> handler from running for cpu changes. This would break PPC, for
> example, which needs to update other information besides the
> elfcorehdr, on cpu changes.
>
> Signed-off-by: Eric DeVolder <eric.devolder@oracle.com>
> ---
>   arch/x86/kernel/crash.c | 10 ++++++++++
>   1 file changed, 10 insertions(+)
>
> diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
> index c9aaec9de775..82ea2b1bdc61 100644
> --- a/arch/x86/kernel/crash.c
> +++ b/arch/x86/kernel/crash.c
> @@ -471,6 +471,16 @@ void arch_crash_handle_hotplug_event(struct kimage *image)
>   	unsigned long mem, memsz;
>   	unsigned long elfsz = 0;
>   
> +	/* As crash_prepare_elf64_headers() has already described all
> +	 * possible cpus, there is no need to update the elfcorehdr
> +	 * for additional cpu changes. This works for both kexec_load()
> +	 * and kexec_file_load() syscalls.
> +	 */
> +	if ((image->file_mode || image->elfcorehdr_updated) &&
> +		((image->hp_action == KEXEC_CRASH_HP_ADD_CPU) ||
> +		(image->hp_action == KEXEC_CRASH_HP_REMOVE_CPU)))
> +		return;

I like the idea of having elfcorehdr_updated to track elfcoredhr segment 
updates.
It handles the possible CPU thing for kexec_load syscall nicely.

Thanks for incorporating the possible CPU change.

The patch series looks good to me.
Reviewed-by: Sourabh Jain <sourabhjain@linux.ibm.com>

- Sourabh Jain
Eric DeVolder March 17, 2023, 6:42 p.m. UTC | #2 
On 3/7/23 03:00, Sourabh Jain wrote:
> 
> On 06/03/23 21:52, Eric DeVolder wrote:
>> This patch is dependent upon the patch 'crash: change
>> crash_prepare_elf64_headers() to for_each_possible_cpu()'. With that
>> patch, crash_prepare_elf64_headers() writes out an ELF cpu PT_NOTE
>> for all possible cpus, thus further cpu changes to the elfcorehdr
>> are not needed.
>>
>> This change works for kexec_file_load() and kexec_load() syscalls.
>> For kexec_file_load(), crash_prepare_elf64_headers() is utilized
>> directly and thus all ELF cpu PT_NOTEs are in the elfcorehdr already.
>> This is the kimage->file_mode term.
>> For kexec_load() syscall, one cpu or memory change will cause the
>> elfcorehdr to be updated via crash_prepare_elf64_headers() and at
>> that point all ELF cpu PT_NOTEs are in the elfcorehdr. This is the
>> kimage->elfcorehdr_updated term.
>>
>> This code is intentionally *NOT* hoisted into
>> crash_handle_hotplug_event() as it would prevent the arch-specific
>> handler from running for cpu changes. This would break PPC, for
>> example, which needs to update other information besides the
>> elfcorehdr, on cpu changes.
>>
>> Signed-off-by: Eric DeVolder <eric.devolder@oracle.com>
>> ---
>>   arch/x86/kernel/crash.c | 10 ++++++++++
>>   1 file changed, 10 insertions(+)
>>
>> diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
>> index c9aaec9de775..82ea2b1bdc61 100644
>> --- a/arch/x86/kernel/crash.c
>> +++ b/arch/x86/kernel/crash.c
>> @@ -471,6 +471,16 @@ void arch_crash_handle_hotplug_event(struct kimage *image)
>>       unsigned long mem, memsz;
>>       unsigned long elfsz = 0;
>> +    /* As crash_prepare_elf64_headers() has already described all
>> +     * possible cpus, there is no need to update the elfcorehdr
>> +     * for additional cpu changes. This works for both kexec_load()
>> +     * and kexec_file_load() syscalls.
>> +     */
>> +    if ((image->file_mode || image->elfcorehdr_updated) &&
>> +        ((image->hp_action == KEXEC_CRASH_HP_ADD_CPU) ||
>> +        (image->hp_action == KEXEC_CRASH_HP_REMOVE_CPU)))
>> +        return;
> 
> I like the idea of having elfcorehdr_updated to track elfcoredhr segment updates.
> It handles the possible CPU thing for kexec_load syscall nicely.
> 
> Thanks for incorporating the possible CPU change.
> 
> The patch series looks good to me.
> Reviewed-by: Sourabh Jain <sourabhjain@linux.ibm.com>
> 
> - Sourabh Jain
> 
Thanks for the RB! Look for v20 very soon.
eric
diff mbox series

Patch

diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index c9aaec9de775..82ea2b1bdc61 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -471,6 +471,16 @@  void arch_crash_handle_hotplug_event(struct kimage *image)
 	unsigned long mem, memsz;
 	unsigned long elfsz = 0;
 
+	/* As crash_prepare_elf64_headers() has already described all
+	 * possible cpus, there is no need to update the elfcorehdr
+	 * for additional cpu changes. This works for both kexec_load()
+	 * and kexec_file_load() syscalls.
+	 */
+	if ((image->file_mode || image->elfcorehdr_updated) &&
+		((image->hp_action == KEXEC_CRASH_HP_ADD_CPU) ||
+		(image->hp_action == KEXEC_CRASH_HP_REMOVE_CPU)))
+		return;
+
 	/*
 	 * Create the new elfcorehdr reflecting the changes to CPU and/or
 	 * memory resources.