[PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[Alan Hayward]

Max size set to 64bits, which I determined using alpha_register_type()

Tested on a --enable-targets=all build using make check with board files
unix and native-gdbserver.

I do not have an ALPHA machine to test on.

Ok to commit?

Alan.


2017-04-04  Alan Hayward  <alan.hayward@arm.com>

	* alpha-tdep.c (alpha_register_to_value): Use ALPHA_MAX_REGISTER_SIZE
	(alpha_value_to_register): Likewise.
	* alpha-tdep.h (ALPHA_MAX_REGISTER_SIZE): Add.



diff --git a/gdb/alpha-tdep.h b/gdb/alpha-tdep.h
index ab46ba56bef1ffb1cea541c506ffed6a5ae9b556..316d94d0bb53e9bc7943d406bf81f7d597437ff3 100644
--- a/gdb/alpha-tdep.h
+++ b/gdb/alpha-tdep.h
@@ -26,6 +26,9 @@ struct regcache;
    real way to know how big a register is.  */
 #define ALPHA_REGISTER_SIZE 8

+/* Big enough to hold the size of the largest register in bytes.  */
+#define ALPHA_MAX_REGISTER_SIZE	8
+
 /* Number of machine registers.  */
 #define ALPHA_NUM_REGS 67

diff --git a/gdb/alpha-tdep.c b/gdb/alpha-tdep.c
index 4dd65c58cc2896ea6038399128e3eb376d1ee672..5a1024e8697a14963b226d275c19721cb6d4e327 100644
--- a/gdb/alpha-tdep.c
+++ b/gdb/alpha-tdep.c
@@ -244,7 +244,7 @@ alpha_register_to_value (struct frame_info *frame, int regnum,
 			int *optimizedp, int *unavailablep)
 {
   struct gdbarch *gdbarch = get_frame_arch (frame);
-  gdb_byte in[MAX_REGISTER_SIZE];
+  gdb_byte in[ALPHA_MAX_REGISTER_SIZE];

   /* Convert to TYPE.  */
   if (!get_frame_register_bytes (frame, regnum, 0,
@@ -266,7 +266,7 @@ static void
 alpha_value_to_register (struct frame_info *frame, int regnum,
 			 struct type *valtype, const gdb_byte *in)
 {
-  gdb_byte out[MAX_REGISTER_SIZE];
+  gdb_byte out[ALPHA_MAX_REGISTER_SIZE];

   switch (TYPE_LENGTH (valtype))
     {

Re: [PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[Yao Qi]

Alan Hayward <Alan.Hayward@arm.com> writes:

> diff --git a/gdb/alpha-tdep.h b/gdb/alpha-tdep.h
> index ab46ba56bef1ffb1cea541c506ffed6a5ae9b556..316d94d0bb53e9bc7943d406bf81f7d597437ff3 100644
> --- a/gdb/alpha-tdep.h
> +++ b/gdb/alpha-tdep.h
> @@ -26,6 +26,9 @@ struct regcache;
>     real way to know how big a register is.  */
>  #define ALPHA_REGISTER_SIZE 8
>
> +/* Big enough to hold the size of the largest register in bytes.  */
> +#define ALPHA_MAX_REGISTER_SIZE	8
> +

> --- a/gdb/alpha-tdep.c
> +++ b/gdb/alpha-tdep.c
> @@ -244,7 +244,7 @@ alpha_register_to_value (struct frame_info *frame, int regnum,
>  			int *optimizedp, int *unavailablep)
>  {
>    struct gdbarch *gdbarch = get_frame_arch (frame);
> -  gdb_byte in[MAX_REGISTER_SIZE];
> +  gdb_byte in[ALPHA_MAX_REGISTER_SIZE];
>
>    /* Convert to TYPE.  */
>    if (!get_frame_register_bytes (frame, regnum, 0,

We can use get_frame_register_value, so "in" can be removed.

> @@ -266,7 +266,7 @@ static void
>  alpha_value_to_register (struct frame_info *frame, int regnum,
>  			 struct type *valtype, const gdb_byte *in)
>  {
> -  gdb_byte out[MAX_REGISTER_SIZE];
> +  gdb_byte out[ALPHA_MAX_REGISTER_SIZE];

Why don't you use ALPHA_REGISTER_SIZE?

-- 
Yao (齐尧)

Re: [PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[Alan Hayward]

> On 11 Apr 2017, at 10:16, Yao Qi <qiyaoltc@gmail.com> wrote:
> 
> Alan Hayward <Alan.Hayward@arm.com> writes:
> 
>> diff --git a/gdb/alpha-tdep.h b/gdb/alpha-tdep.h
>> index ab46ba56bef1ffb1cea541c506ffed6a5ae9b556..316d94d0bb53e9bc7943d406bf81f7d597437ff3 100644
>> --- a/gdb/alpha-tdep.h
>> +++ b/gdb/alpha-tdep.h
>> @@ -26,6 +26,9 @@ struct regcache;
>>    real way to know how big a register is.  */
>> #define ALPHA_REGISTER_SIZE 8
>> 
>> +/* Big enough to hold the size of the largest register in bytes.  */
>> +#define ALPHA_MAX_REGISTER_SIZE	8
>> +
> 
>> --- a/gdb/alpha-tdep.c
>> +++ b/gdb/alpha-tdep.c
>> @@ -244,7 +244,7 @@ alpha_register_to_value (struct frame_info *frame, int regnum,
>> 			int *optimizedp, int *unavailablep)
>> {
>>   struct gdbarch *gdbarch = get_frame_arch (frame);
>> -  gdb_byte in[MAX_REGISTER_SIZE];
>> +  gdb_byte in[ALPHA_MAX_REGISTER_SIZE];
>> 
>>   /* Convert to TYPE.  */
>>   if (!get_frame_register_bytes (frame, regnum, 0,
> 
> We can use get_frame_register_value, so "in" can be removed.
> 
>> @@ -266,7 +266,7 @@ static void
>> alpha_value_to_register (struct frame_info *frame, int regnum,
>> 			 struct type *valtype, const gdb_byte *in)
>> {
>> -  gdb_byte out[MAX_REGISTER_SIZE];
>> +  gdb_byte out[ALPHA_MAX_REGISTER_SIZE];
> 
> Why don't you use ALPHA_REGISTER_SIZE?

Hmmm... yes. Not sure why I ignored that!

> 
> -- 
> Yao (齐尧)


Updated changes below.

I don't have a ALPHA machine to test on.
Tested on a --enable-targets=all build using make check with board files
unix and native-gdbserver.

Ok to commit?

Alan.

2017-04-12  Alan Hayward  <alan.hayward@arm.com>

	* alpha-tdep.c (alpha_register_to_value): Use get_frame_register_value.
	(alpha_value_to_register): Use ALPHA_REGISTER_SIZE.

diff --git a/gdb/alpha-tdep.c b/gdb/alpha-tdep.c
index 4dd65c58cc2896ea6038399128e3eb376d1ee672..2f9f81952fea0e7fa050866db3fe05776edcdc63 100644
--- a/gdb/alpha-tdep.c
+++ b/gdb/alpha-tdep.c
@@ -244,17 +244,23 @@ alpha_register_to_value (struct frame_info *frame, int regnum,
 			int *optimizedp, int *unavailablep)
 {
   struct gdbarch *gdbarch = get_frame_arch (frame);
-  gdb_byte in[MAX_REGISTER_SIZE];
+  struct value *value;

   /* Convert to TYPE.  */
-  if (!get_frame_register_bytes (frame, regnum, 0,
-				 register_size (gdbarch, regnum),
-				 in, optimizedp, unavailablep))
+
+  value = get_frame_register_value (frame, regnum);
+
+  gdb_assert (value != NULL);
+
+  *optimizedp = value_optimized_out (value);
+  *unavailablep = !value_entirely_available (value);
+
+  if (*optimizedp || *unavailablep)
     return 0;

   if (TYPE_LENGTH (valtype) == 4)
     {
-      alpha_sts (gdbarch, out, in);
+      alpha_sts (gdbarch, out, value_contents (value));
       *optimizedp = *unavailablep = 0;
       return 1;
     }
@@ -266,7 +272,7 @@ static void
 alpha_value_to_register (struct frame_info *frame, int regnum,
 			 struct type *valtype, const gdb_byte *in)
 {
-  gdb_byte out[MAX_REGISTER_SIZE];
+  gdb_byte out[ALPHA_REGISTER_SIZE];

   switch (TYPE_LENGTH (valtype))
     {

Re: [PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[Alan Hayward]

> On 12 Apr 2017, at 09:25, Alan Hayward <Alan.Hayward@arm.com> wrote:
> 
>> 
>> On 11 Apr 2017, at 10:16, Yao Qi <qiyaoltc@gmail.com> wrote:
>> 
>> Alan Hayward <Alan.Hayward@arm.com> writes:
>> 

I’ve rebased this patch due to Yao’s unit test changes.
Added two asserts a well.


I don't have a ALPHA machine to test on.
Tested on a --enable-targets=all and --enable-libsanitizer build using
make check with board files unix and native-gdbserver.

Ok to commit?

Alan.

2017-05-30  Alan Hayward  <alan.hayward@arm.com>

	* alpha-tdep.c (alpha_register_to_value): Use get_frame_register_value.
	(alpha_value_to_register): Use ALPHA_REGISTER_SIZE.


diff --git a/gdb/alpha-tdep.c b/gdb/alpha-tdep.c
index d7cc0f4974362af89c2b42f4d5934fa8cd5d8a39..06ee90cbc7785d8277e773bdb6b98065e2e19568 100644
--- a/gdb/alpha-tdep.c
+++ b/gdb/alpha-tdep.c
@@ -244,17 +244,26 @@ alpha_register_to_value (struct frame_info *frame, int regnum,
 			int *optimizedp, int *unavailablep)
 {
   struct gdbarch *gdbarch = get_frame_arch (frame);
-  gdb_byte in[MAX_REGISTER_SIZE];
+  struct value *value = get_frame_register_value (frame, regnum);

-  /* Convert to TYPE.  */
-  if (!get_frame_register_bytes (frame, regnum, 0,
-				 register_size (gdbarch, regnum),
-				 in, optimizedp, unavailablep))
-    return 0;
+  gdb_assert (value != NULL);
+  *optimizedp = value_optimized_out (value);
+  *unavailablep = !value_entirely_available (value);
+
+  if (*optimizedp || *unavailablep)
+    {
+      release_value (value);
+      value_free (value);
+      return 0;
+    }
+
+  /* Convert to VALTYPE.  */

   gdb_assert (TYPE_LENGTH (valtype) == 4);
-  alpha_sts (gdbarch, out, in);
-  *optimizedp = *unavailablep = 0;
+  alpha_sts (gdbarch, out, value_contents_all (value));
+
+  release_value (value);
+  value_free (value);
   return 1;
 }

@@ -262,9 +271,11 @@ static void
 alpha_value_to_register (struct frame_info *frame, int regnum,
 			 struct type *valtype, const gdb_byte *in)
 {
-  gdb_byte out[MAX_REGISTER_SIZE];
+  gdb_byte out[ALPHA_REGISTER_SIZE];

   gdb_assert (TYPE_LENGTH (valtype) == 4);
+  gdb_assert (register_size (get_frame_arch (frame), regnum)
+	      <= ALPHA_REGISTER_SIZE);
   alpha_lds (get_frame_arch (frame), out, in);

   put_frame_register (frame, regnum, out);

Re: [PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[Yao Qi]

Alan Hayward <Alan.Hayward@arm.com> writes:

> I don't have a ALPHA machine to test on.
> Tested on a --enable-targets=all and --enable-libsanitizer build using
> make check with board files unix and native-gdbserver.

Could you build GDB with option -fsanitize=address
(make CXXFLAGS='-fsanitize=address') and run gdb.gdb/unittest.exp?

>
> Ok to commit?
>

OK if there is no fail in unit tests.

> Alan.
>
> 2017-05-30  Alan Hayward  <alan.hayward@arm.com>
>
> 	* alpha-tdep.c (alpha_register_to_value): Use get_frame_register_value.

This line is too long.  The line max length in ChangeLog is 74.
https://sourceware.org/gdb/wiki/ContributionChecklist#Properly_Formatted_GNU_ChangeLog

> 	(alpha_value_to_register): Use ALPHA_REGISTER_SIZE.

-- 
Yao (齐尧)

Re: [PATCH 6/11] Add ALPHA_MAX_REGISTER_SIZE

[Alan Hayward]

> On 6 Jun 2017, at 15:02, Yao Qi <qiyaoltc@gmail.com> wrote:
> 
> Alan Hayward <Alan.Hayward@arm.com> writes:
> 
>> I don't have a ALPHA machine to test on.
>> Tested on a --enable-targets=all and --enable-libsanitizer build using
>> make check with board files unix and native-gdbserver.
> 
> Could you build GDB with option -fsanitize=address
> (make CXXFLAGS='-fsanitize=address') and run gdb.gdb/unittest.exp?
> 
>> 
>> Ok to commit?
>> 
> 
> OK if there is no fail in unit tests.

Tested along with the IA64 patch.
unittest.exp passes.


> 
>> Alan.
>> 
>> 2017-05-30  Alan Hayward  <alan.hayward@arm.com>
>> 
>> 	* alpha-tdep.c (alpha_register_to_value): Use get_frame_register_value.
> 
> This line is too long.  The line max length in ChangeLog is 74.
> https://sourceware.org/gdb/wiki/ContributionChecklist#Properly_Formatted_GNU_ChangeLog

Sorry, that was new to me (As far as I’m aware, gcc uses 80).


> 
>> 	(alpha_value_to_register): Use ALPHA_REGISTER_SIZE.
> 
> 

Committed with newline added in changelog.


Alan.