Patchwork [AArch64,5/6] Implement gdbarch_core_read_description

login
register
mail settings
Submitter Jiong Wang
Date Aug. 9, 2017, 3:43 p.m.
Message ID <5b77a77e-e0e0-a420-6f8e-ae8c3e967d68@foss.arm.com>
Download mbox | patch
Permalink /patch/22047/
State New
Headers show

Comments

Jiong Wang - Aug. 9, 2017, 3:43 p.m.
On 09/08/17 15:57, Jiong Wang wrote:
>
>> This patch doesn't depend on other patches, so it can go in now.
>>
> I had committed this patch after addressing all comments, but just reverted.
>
> it's my bad haven't give it a quick build.  It depends on AT_HWCAP defined in patch 1 and the value needs to be synced with kernel...
>

Sorry, the issue should be two header files are missing.

"auxv.h" to declare target_auxv_search and "elf/common.h" to define AT_HWCAP
(I miss read the buildbot error message as something like missing "HWCAP_APIA"...)

This patch was splitted from patch 6, I forget to move those two header files to
this patch.

Attached patch fixed this and passed gdb cross build, OK for master?

gdb/
2017-08-09  Jiong Wang  <jiong.wang@arm.com>

	* aarch64-linux-tdep.c: Include "auxv.h" and "elf/common.h".
	(aarch64_linux_core_read_description): New function.
	(aarch64_linux_init_abi): Register gdbarch_core_read_description.
Yao Qi - Aug. 9, 2017, 4:03 p.m.
Jiong Wang <jiong.wang@foss.arm.com> writes:

> "auxv.h" to declare target_auxv_search and "elf/common.h" to define AT_HWCAP
> (I miss read the buildbot error message as something like missing
> "HWCAP_APIA"...)
>

Is it because the buildbot error message is too scary? :)

> This patch was splitted from patch 6, I forget to move those two header files to
> this patch.
>
> Attached patch fixed this and passed gdb cross build, OK for master?

Yes, please.
Jiong Wang - Aug. 9, 2017, 4:40 p.m.
On 09/08/17 17:03, Yao Qi wrote:

> Jiong Wang <jiong.wang@foss.arm.com> writes:
>
>> "auxv.h" to declare target_auxv_search and "elf/common.h" to define AT_HWCAP
>> (I miss read the buildbot error message as something like missing
>> "HWCAP_APIA"...)
>>
> Is it because the buildbot error message is too scary? :)

I was chuckle to myself after pushing the patch, the suddenly pop-up of buildbot
error message with "unfortunate" in the summary let me feel I got a sentence and
was nervously hurried to scroll down to the bottom, oh, my god, it's lifetime...


>
>> This patch was splitted from patch 6, I forget to move those two header files to
>> this patch.
>>
>> Attached patch fixed this and passed gdb cross build, OK for master?
> Yes, please.
>

Patch

diff --git a/gdb/aarch64-linux-tdep.c b/gdb/aarch64-linux-tdep.c
index d2ca70a..847454f 100644
--- a/gdb/aarch64-linux-tdep.c
+++ b/gdb/aarch64-linux-tdep.c
@@ -45,6 +45,8 @@ 
 
 #include "record-full.h"
 #include "linux-record.h"
+#include "auxv.h"
+#include "elf/common.h"
 
 /* Signal frame handling.
 
@@ -231,6 +233,20 @@  aarch64_linux_iterate_over_regset_sections (struct gdbarch *gdbarch,
       NULL, cb_data);
 }
 
+/* Implement the "core_read_description" gdbarch method.  */
+
+static const struct target_desc *
+aarch64_linux_core_read_description (struct gdbarch *gdbarch,
+				     struct target_ops *target, bfd *abfd)
+{
+  CORE_ADDR aarch64_hwcap = 0;
+
+  if (target_auxv_search (target, AT_HWCAP, &aarch64_hwcap) != 1)
+    return NULL;
+
+  return tdesc_aarch64;
+}
+
 /* Implementation of `gdbarch_stap_is_single_operand', as defined in
    gdbarch.h.  */
 
@@ -1018,6 +1034,8 @@  aarch64_linux_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch)
 
   set_gdbarch_iterate_over_regset_sections
     (gdbarch, aarch64_linux_iterate_over_regset_sections);
+  set_gdbarch_core_read_description
+    (gdbarch, aarch64_linux_core_read_description);
 
   /* SystemTap related.  */
   set_gdbarch_stap_integer_prefixes (gdbarch, stap_integer_prefixes);