Luke Craig <lacra...@gmail.com> writes:
> ---
> include/qemu/qemu-plugin.h | 11 +++++++++++
> plugins/api.c | 13 +++++++++++++
> 2 files changed, 24 insertions(+)
>
> diff --git a/include/qemu/qemu-plugin.h b/include/qemu/qemu-plugin.h
> index 3a850aa216..a1c478c54f 100644
> --- a/include/qemu/qemu-plugin.h
> +++ b/include/qemu/qemu-plugin.h
> @@ -500,6 +500,17 @@ QEMU_PLUGIN_API
> struct qemu_plugin_insn *
> qemu_plugin_tb_get_insn(const struct qemu_plugin_tb *tb, size_t idx);
>
> +/**
> + * qemu_plugin_tb_get_insn_by_vaddr() - lookup handle for instruction by
> vaddr
> + * @tb: opaque handle to TB passed to callback
> + * @vaddr: virtual address of instruction
> + *
> + * Returns: opaque handle to instruction
> + */
> +QEMU_PLUGIN_API
> +struct qemu_plugin_insn *
> +qemu_plugin_tb_get_insn_by_vaddr(const struct qemu_plugin_tb *tb, uint64_t
> vaddr);
> +
> /**
> * qemu_plugin_insn_data() - copy instruction data
> * @insn: opaque instruction handle from qemu_plugin_tb_get_insn()
> diff --git a/plugins/api.c b/plugins/api.c
> index 4110cfaa23..7ff5e1c1bd 100644
> --- a/plugins/api.c
> +++ b/plugins/api.c
> @@ -258,6 +258,19 @@ qemu_plugin_tb_get_insn(const struct qemu_plugin_tb *tb,
> size_t idx)
> return insn;
> }
>
> +struct qemu_plugin_insn *
> +qemu_plugin_tb_get_insn_by_vaddr(const struct qemu_plugin_tb *tb, uint64_t
> vaddr)
> +{
> + struct qemu_plugin_insn *insn;
> + for (size_t i = 0; i < tb->n; i++){
> + insn = g_ptr_array_index(tb->insns, idx);
> + if (insn != NULL && insn->vaddr == vaddr){
> + return insn;
> + }
> + }
> + return NULL;
> +}
> +
I don't have any fundamental objection to this. I would prefer it if the
new helper was used by either one of the test or contrib plugins to
ensure we don't bitrot it.
> /*
> * Instruction information
> *
--
Alex Bennée
Virtualisation Tech Lead @ Linaro
