On 1 July 2015 at 12:00, Bhushan Attarde <bhushan.atta...@imgtec.com> wrote:
> Implementation of "Xfer:auxv:read" to provide auxiliary vector information
> to clients which relies on it.
>
> For example: AT_ENTRY in auxiliary vector provides the entry point
> information.
> Client can use this information to compare it with entry point mentioned in
> executable to calculate load offset and then update the load addresses
> accordingly.
>
> Signed-off-by: Bhushan Attarde <bhushan.atta...@imgtec.com>
> ---
> gdbstub.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 47 insertions(+)
>
> diff --git a/gdbstub.c b/gdbstub.c
> index cea2a84..bdcaa6b 100644
> --- a/gdbstub.c
> +++ b/gdbstub.c
> @@ -1153,6 +1153,9 @@ static int gdb_handle_packet(GDBState *s, const char
> *line_buf)
> if (cc->gdb_core_xml_file != NULL) {
> pstrcat(buf, sizeof(buf), ";qXfer:features:read+");
> }
> +#ifdef CONFIG_USER_ONLY
> + pstrcat(buf, sizeof(buf), ";qXfer:auxv:read+");
> +#endif
> put_packet(s, buf);
> break;
> }
> @@ -1199,6 +1202,50 @@ static int gdb_handle_packet(GDBState *s, const char
> *line_buf)
> put_packet_binary(s, buf, len + 1);
> break;
> }
> +#ifdef CONFIG_USER_ONLY
> + if (strncmp(p, "Xfer:auxv:read:", 15) == 0) {
> + TaskState *ts = s->c_cpu->opaque;
> + target_ulong auxv = ts->info->saved_auxv;
> + target_ulong auxv_len = ts->info->auxv_len;
> + char *ptr;
> +
> + p += 15;
> + while (*p && *p != ':') {
> + p++;
> + }
> + p++;
> +
> + addr = strtoul(p, (char **)&p, 16);
> + if (*p == ',') {
> + p++;
> + }
> + len = strtoul(p, (char **)&p, 16);
> +
> + ptr = lock_user(VERIFY_READ, auxv, auxv_len, 0);
> + if (ptr == NULL) {
> + break;
> + }
> +
> + if (addr > len) {
> + snprintf(buf, sizeof(buf), "E00");
> + put_packet(s, buf);
> + break;
> + }
> + if (len > (MAX_PACKET_LENGTH - 5) / 2) {
> + len = (MAX_PACKET_LENGTH - 5) / 2;
> + }
> + if (len < auxv_len - addr) {
> + buf[0] = 'm';
> + len = memtox(buf + 1, ptr + addr, len);
> + } else {
> + buf[0] = 'l';
> + len = memtox(buf + 1, ptr + addr, auxv_len - addr);
> + }
The bulk of this code for parsing out the arguments in a
qXfer:OBJECT:read:ANNEX:OFFSET,LENGTH packet is identical
to that we already have for handling Xfer:features:read.
We should factor this out neatly rather than duplicating it.
thanks
-- PMM
