Sure, Thanks for remind me of this.
One thing I must point out, in current code logic, if "START_SEG_ADDR_RECORD"
occured multiple times, only the last one works. I don't know whether GNU
binutils would emit 'The Record' many times.
Best,
SU Hang
> -----Original Messages-----
> From: "Stefan Hajnoczi" <stefa...@redhat.com>
> Sent Time: 2018-06-10 00:32:52 (Sunday)
> To: qemu-devel@nongnu.org
> Cc: "Su Hang" <suhan...@mails.ucas.ac.cn>, jus...@mail.ru,
> j...@groklearning.com, j...@jms.id.au, qemu...@steffen-goertz.de, "Stefan
> Hajnoczi" <stefa...@redhat.com>
> Subject: [PATCH] loader: implement START_SEG_ADDR_RECORD
>
> It turns out that GNU binutils emits START_SEG_ADDR_RECORD when the start
> address is within the first megabyte (< 0x100000). Therefore we must
> handle this record type.
>
> Originally we thought this record type was x86-specific, but binutils
> also emits it on non-x86 architectures.
>
> Based-on: <1527161340-3200-1-git-send-email-suhan...@mails.ucas.ac.cn>
> Cc: Su Hang <suhan...@mails.ucas.ac.cn>
> Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
> ---
> Su Hang: Feel free to squash this into the next revision of your hex
> loader patch. Don't worry about the authorship information.
>
> hw/core/loader.c | 10 ++++++++--
> 1 file changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/hw/core/loader.c b/hw/core/loader.c
> index 3c0202caa8..7843b487b2 100644
> --- a/hw/core/loader.c
> +++ b/hw/core/loader.c
> @@ -1423,8 +1423,14 @@ static int handle_record_type(HexParser *parser)
> break;
>
> case START_SEG_ADDR_RECORD:
> - /* TODO: START_SEG_ADDR_RECORD is x86-specific */
> - return -1;
> + if (line->byte_count != 4 && line->address != 0) {
> + return -1;
> + }
> +
> + /* x86 16-bit CS:IP segmented addressing */
> + *(parser->start_addr) = (((line->data[0] << 8) | line->data[1]) <<
> 4) |
> + (line->data[2] << 8) | line->data[3];
> + break;
>
> case START_LINEAR_ADDR_RECORD:
> if (line->byte_count != 4 && line->address != 0) {
> --
> 2.17.1
