On 7/26/22 11:22, Janosch Frank wrote:
> By splitting the writing of the section headers and (future) section
> data we prepare for the addition of a string table section and
> architecture sections.
>
> At the same time we move the writing of the section to the end of the
> dump process. This allows the upcoming architecture section code to
> add data after all of the common dump data has been written.
>
> Signed-off-by: Janosch Frank <fran...@linux.ibm.com>
> ---
> dump/dump.c | 112 ++++++++++++++++++++++++++++++++----------
> include/sysemu/dump.h | 4 ++
> 2 files changed, 90 insertions(+), 26 deletions(-)
>
> diff --git a/dump/dump.c b/dump/dump.c
> index 8a7ce95610..a6bb7bfa21 100644
> --- a/dump/dump.c
> +++ b/dump/dump.c
> @@ -342,30 +342,69 @@ static void write_elf_phdr_note(DumpState *s, Error
> **errp)
> }
> }
>
[...]
> +
> +static void prepare_elf_section_hdrs(DumpState *s)
> +{
> + size_t len, sizeof_shdr;
> +
> + /*
> + * Section ordering:
> + * - HDR zero (if needed)
> + */
> + sizeof_shdr = dump_is_64bit(s) ? sizeof(Elf64_Shdr) : sizeof(Elf32_Shdr);
> + len = sizeof_shdr * s->shdr_num;
> + s->elf_section_hdrs = g_malloc0(len);
> +
> + /* Write special section first */
> + if (s->phdr_num == PN_XNUM) {
Should be >= right?
> + prepare_elf_section_hdr_zero(s);
> + }
> +}
> +
[...]
