On 2024/10/09 15:06, Ming Wang wrote:
> The calculation of the starting PFN for exclusion in
> exclude_nodata_pages() was rounding down, potentially excluding
> the last valid page of a LOAD segment.
>
> This commit fixes the issue by rounding up the calculated PFN to the
> nearest page boundary, ensuring that the last valid page is included and
> not erroneously excluded.
>
> Signed-off-by: Ming Wang <wangming01@xxxxxxxxxxx>
> ---
> makedumpfile.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/makedumpfile.c b/makedumpfile.c
> index 2fa67e7..b356eb3 100644
> --- a/makedumpfile.c
> +++ b/makedumpfile.c
> @@ -5215,7 +5215,7 @@ exclude_nodata_pages(struct cycle *cycle)
> NULL, &file_size)) {
> unsigned long long pfn, pfn_end;
>
> - pfn = paddr_to_pfn(phys_start + file_size);
> + pfn = paddr_to_pfn(roundup(phys_start + file_size, PAGESIZE()));
> pfn_end = paddr_to_pfn(roundup(phys_end, PAGESIZE()));
>
> if (pfn < cycle->start_pfn)
Thank you for the v2 patch.
I confirmed that without the patch "makedumpfile -d 0" excluded some pages wrongly:
$ makedumpfile -d 0 vmcore dump.d0
...
pfn 0000000000054000 pfn_end 0000000000054000
pfn 000000000006989f pfn_end 00000000000698a0
crash> ptob 6989f
6989f: 6989f000
crash> rd -p 6989f000
rd: page excluded: physical address: 6989f000 type: "64-bit PHYSADDR"
and it didn't with the patch:
$ makedumpfile -d 0 vmcore dump.d0.patched
...
pfn 0000000000054000 pfn_end 0000000000054000
pfn 00000000000698a0 pfn_end 00000000000698a0
crash> rd -p 6989f000
6989f000: cccccccc30646870 phd0....
Tested ok and applied.
https://github.com/makedumpfile/makedumpfile/commit/f2c5f761ea99dbacaaf988caf1d51bc142b353fa
Thanks,
Kazu
_______________________________________________
kexec mailing list
kexec@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/kexec
