On 09/25/2013 04:39 PM, HATAYAMA Daisuke wrote: > (2013/09/25 17:16), Jingbai Ma wrote: > <cut> >>> >>>> @@ -5153,10 +5160,11 @@ write_kdump_header(void) >>>> * Write common header >>>> */ >>>> strncpy(dh->signature, KDUMP_SIGNATURE, strlen(KDUMP_SIGNATURE)); >>>> - dh->header_version = 5; >>>> + dh->header_version = 6; >>>> dh->block_size = info->page_size; >>>> dh->sub_hdr_size = sizeof(kh) + size_note; >>>> dh->sub_hdr_size = divideup(dh->sub_hdr_size, dh->block_size); >>>> + /* dh->max_mapnr may be truncated here, full 64bit in kh.max_mapnr */ >>>> dh->max_mapnr = info->max_mapnr; >>> >>> dh->max_mapnr = MIN(info->max_mapnr, UINT_MAX) seems better for old >>> versions of crash utitlity. >>> >> >> Although change this value to UINT_MAX doesn't help the old crash >> utility very much. At least this special value will tell the user >> something happened. >> Will fix it. >> >> > > This is the largest page frame number old crash utilities can represent, > not special value. > By this, old crash utilities can read at least the pages less than > UINT_MAX, > which is better than making dh->max_mapnr overflow; then dh->max_mapnr > would typically become > a small number. > I knew, but crash utility may still report reading page error before its command prompt. Anyway, UINT_MAX has a better chance to make old utility happy, so I will fix it. :) -- Thanks, Jingbai Ma
