[Qemu-devel] [PATCH V3] gdbstub: Fix handler for 'F' packet

Sandra Loosemore posted 1 patch 6 years, 1 month ago
Test docker-clang@ubuntu passed
Test FreeBSD passed
Test checkpatch passed
Test docker-mingw@fedora passed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20190827223317.8614-1-sandra@codesourcery.com
Maintainers: "Philippe Mathieu-Daudé" <philmd@redhat.com>, "Alex Bennée" <alex.bennee@linaro.org>
gdbstub.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
[Qemu-devel] [PATCH V3] gdbstub: Fix handler for 'F' packet
Posted by Sandra Loosemore 6 years, 1 month ago
Handling of the 'F' packet has been broken since commit
4b20fab101b9e2d0fb47454209637a17fc7a13d5, which converted it to use
the new packet parsing infrastructure.  Per the GDB RSP specification

https://sourceware.org/gdb/current/onlinedocs/gdb/The-F-Reply-Packet.html

the second parameter may be omitted, but the rewritten implementation
was failing to recognize this case.  The result was that QEMU was
repeatedly resending the fileio request and ignoring GDB's replies of
successful completion.  This patch restores the behavior of the
previous code in allowing the errno parameter to be omitted and
passing 0 to the callback in that case.

Signed-off-by: Sandra Loosemore <sandra@codesourcery.com>
---
 gdbstub.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/gdbstub.c b/gdbstub.c
index b92ba59..3e8bcd0 100644
--- a/gdbstub.c
+++ b/gdbstub.c
@@ -1820,11 +1820,15 @@ static void handle_read_all_regs(GdbCmdContext *gdb_ctx, void *user_ctx)
 
 static void handle_file_io(GdbCmdContext *gdb_ctx, void *user_ctx)
 {
-    if (gdb_ctx->num_params >= 2 && gdb_ctx->s->current_syscall_cb) {
+    if (gdb_ctx->num_params >= 1 && gdb_ctx->s->current_syscall_cb) {
         target_ulong ret, err;
 
         ret = (target_ulong)gdb_ctx->params[0].val_ull;
-        err = (target_ulong)gdb_ctx->params[1].val_ull;
+        if (gdb_ctx->num_params >= 2) {
+            err = (target_ulong)gdb_ctx->params[1].val_ull;
+        } else {
+            err = 0;
+        }
         gdb_ctx->s->current_syscall_cb(gdb_ctx->s->c_cpu, ret, err);
         gdb_ctx->s->current_syscall_cb = NULL;
     }
-- 
2.8.1


Re: [Qemu-devel] [PATCH V3] gdbstub: Fix handler for 'F' packet
Posted by Richard Henderson 6 years, 1 month ago
On 8/27/19 3:33 PM, Sandra Loosemore wrote:
> Handling of the 'F' packet has been broken since commit
> 4b20fab101b9e2d0fb47454209637a17fc7a13d5, which converted it to use
> the new packet parsing infrastructure.  Per the GDB RSP specification
> 
> https://sourceware.org/gdb/current/onlinedocs/gdb/The-F-Reply-Packet.html
> 
> the second parameter may be omitted, but the rewritten implementation
> was failing to recognize this case.  The result was that QEMU was
> repeatedly resending the fileio request and ignoring GDB's replies of
> successful completion.  This patch restores the behavior of the
> previous code in allowing the errno parameter to be omitted and
> passing 0 to the callback in that case.
> 
> Signed-off-by: Sandra Loosemore <sandra@codesourcery.com>
> ---
>  gdbstub.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~


Re: [Qemu-devel] [PATCH V3] gdbstub: Fix handler for 'F' packet
Posted by Alex Bennée 6 years, 1 month ago
Sandra Loosemore <sandra@codesourcery.com> writes:

> Handling of the 'F' packet has been broken since commit
> 4b20fab101b9e2d0fb47454209637a17fc7a13d5, which converted it to use
> the new packet parsing infrastructure.  Per the GDB RSP specification
>
> https://sourceware.org/gdb/current/onlinedocs/gdb/The-F-Reply-Packet.html
>
> the second parameter may be omitted, but the rewritten implementation
> was failing to recognize this case.  The result was that QEMU was
> repeatedly resending the fileio request and ignoring GDB's replies of
> successful completion.  This patch restores the behavior of the
> previous code in allowing the errno parameter to be omitted and
> passing 0 to the callback in that case.
>
> Signed-off-by: Sandra Loosemore <sandra@codesourcery.com>

Queued to gdbstub/next, thanks.

> ---
>  gdbstub.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/gdbstub.c b/gdbstub.c
> index b92ba59..3e8bcd0 100644
> --- a/gdbstub.c
> +++ b/gdbstub.c
> @@ -1820,11 +1820,15 @@ static void handle_read_all_regs(GdbCmdContext *gdb_ctx, void *user_ctx)
>
>  static void handle_file_io(GdbCmdContext *gdb_ctx, void *user_ctx)
>  {
> -    if (gdb_ctx->num_params >= 2 && gdb_ctx->s->current_syscall_cb) {
> +    if (gdb_ctx->num_params >= 1 && gdb_ctx->s->current_syscall_cb) {
>          target_ulong ret, err;
>
>          ret = (target_ulong)gdb_ctx->params[0].val_ull;
> -        err = (target_ulong)gdb_ctx->params[1].val_ull;
> +        if (gdb_ctx->num_params >= 2) {
> +            err = (target_ulong)gdb_ctx->params[1].val_ull;
> +        } else {
> +            err = 0;
> +        }
>          gdb_ctx->s->current_syscall_cb(gdb_ctx->s->c_cpu, ret, err);
>          gdb_ctx->s->current_syscall_cb = NULL;
>      }


--
Alex Bennée