On 3/16/23 09:44, Eric DeVolder wrote:
> 
> 
> On 3/16/23 05:11, Baoquan He wrote:
>> On 03/06/23 at 11:22am, Eric DeVolder wrote:
>> ......
>>> +static void crash_handle_hotplug_event(unsigned int hp_action, unsigned int cpu)
>>> +{
>>> +    /* Obtain lock while changing crash information */
>>> +    if (kexec_trylock()) {
>>> +
>>> +        /* Check kdump is loaded */
>>> +        if (kexec_crash_image) {
>>> +            struct kimage *image = kexec_crash_image;
>>> +
>>> +            if (hp_action == KEXEC_CRASH_HP_ADD_CPU ||
>>> +                hp_action == KEXEC_CRASH_HP_REMOVE_CPU)
>>> +                pr_debug("hp_action %u, cpu %u\n", hp_action, cpu);
>>> +            else
>>> +                pr_debug("hp_action %u\n", hp_action);
>>> +
>>> +            /*
>>> +             * When the struct kimage is allocated, the elfcorehdr_index
>>> +             * is set to -1. Find the segment containing the elfcorehdr,
>>> +             * if not already found. This works for both the kexec_load
>>> +             * and kexec_file_load paths.
>>> +             */
>>> +            if (image->elfcorehdr_index < 0) {
>>> +                unsigned long mem;
>>> +                unsigned char *ptr;
>>> +                unsigned int n;
>>> +
>>> +                for (n = 0; n < image->nr_segments; n++) {
>>> +                    mem = image->segment[n].mem;
>>> +                    ptr = kmap_local_page(pfn_to_page(mem >> PAGE_SHIFT));
>>> +                    if (ptr) {
>>> +                        /* The segment containing elfcorehdr */
>>> +                        if (memcmp(ptr, ELFMAG, SELFMAG) == 0) {
>>> +                            image->elfcorehdr_index = (int)n;
>>> +                        }
>>> +                        kunmap_local(ptr);
>>> +                    }
>>> +                }
>>> +            }
>>> +
>>> +            if (image->elfcorehdr_index < 0) {
>>> +                pr_err("unable to locate elfcorehdr segment");
>>> +                goto out;
>>> +            }
>>> +
>>> +            /* Needed in order for the segments to be updated */
>>> +            arch_kexec_unprotect_crashkres();
>>> +
>>> +            /* Differentiate between normal load and hotplug update */
>>> +            image->hp_action = hp_action;
>>> +
>>> +            /* Now invoke arch-specific update handler */
>>> +            arch_crash_handle_hotplug_event(image);
>>> +
>>> +            /* No longer handling a hotplug event */
>>> +            image->hp_action = KEXEC_CRASH_HP_NONE;
>>> +            image->elfcorehdr_updated = true;
>>
>> It's good to initialize the image->hp_action here, however where do
>> you check it? Do you plan to add some check somewhere?
> 
> Hi Baoquan,
> The hp_action member is initialized to 0 in do_image_alloc_init(). I've
> mapped KEXEC_CRASH_HP_NONE onto 0 on purpose.
> 
> But the use of image->hp_action = KEXEC_CRASH_HP_NONE is to actually
> delineate that a hotplug event handling has completed. You can see
> imae->hp_action set to hp_action to capture what the triggering event
> was, as passed into this function.
> 
> I will go ahead and set image->hp_action = KEXEC_CRASH_HP_NONE; explicitly
> in do_kimage_alloc_init(), as that is done for the other crash hotplug members.
> 
> Thanks!
> eric

Baoquan, as you are currently working through v19, please let me know when I should put out v20 with 
this and Sourabh's feedback.
Thanks,
eric

> 
> 
>>
>>> +
>>> +            /* Change back to read-only */
>>> +            arch_kexec_protect_crashkres();
>>> +        }
>>> +
>>> +out:
>>> +        /* Release lock now that update complete */
>>> +        kexec_unlock();
>>> +    }
>>> +}
>> ......
>>