Re: [PATCH 3/3] Calculate max register size

[Alan Hayward <Alan.Hayward@arm.com>]

> On 9 Jan 2017, at 20:14, Luis Machado <lgustavo@codesourcery.com> wrote:
> 
> On 01/09/2017 04:58 AM, Alan Hayward wrote:
>> Aarch64 SVE requires a max register size of 256. The current max size in gdb
>> is 64. This is part of a series demonstrating the replacement of
>> MAX_REGISTER_SIZE.
>> 
>> In cases where a buffer is created to be used multiple times to hold different
>> registers, then the maximum register size is required. Add a max register value
>> to the regcache which is calculated on initialization.
>> 
>> This patch is restricted to remote.c and regcache.c.
>> Follow on patches will expand to other files.
>> 
>> Tested on x86.
>> Ok to commit?
>> 
>> Thanks,
>> Alan.
>> 
>> 2017-01-09  Alan Hayward <alan.hayward@arm.com>
>> 
>> 	* regcache.c (struct regcache_descr): Add max_register_size
>> 	(max_register_size): New.
>> 	(init_regcache_descr): Find max register size.
>> 	(regcache_save): Use max_register_size.
>> 	(regcache_restore): Likewise.
>> 	(regcache_dump): Likewise.
>> 	* regcache.h (max_register_size): New.
>> 	* remote.c (remote_prepare_to_store): Allocate buffer.
>> 
>> 
>> diff --git a/gdb/regcache.h b/gdb/regcache.h
>> index e5a7cf553279b8cc0d546ec1b8274cbf97e246d5..4db9517a9dd464d9c43be2af0573b767b86bfb56 100644
>> --- a/gdb/regcache.h
>> +++ b/gdb/regcache.h
>> @@ -202,6 +202,9 @@ extern struct type *register_type (struct gdbarch *gdbarch, int regnum);
>> 
>> extern int register_size (struct gdbarch *gdbarch, int regnum);
>> 
>> +/* Return the size of the largest register.  */
>> +
> 
> Spurious newline.

Ok.

> 
>> +extern long max_register_size (struct gdbarch *gdbarch);
>> 
>> /* Save/restore a register cache.  The set of registers saved /
>>    restored into the DST regcache determined by the save_reggroup /
>> diff --git a/gdb/regcache.c b/gdb/regcache.c
>> index 9d28aa2c2114e0f1c52758bb2fbe9669a329c13e..46d017c7b2abcb18c9cdda005749071328735dbd 100644
>> --- a/gdb/regcache.c
>> +++ b/gdb/regcache.c
>> @@ -73,6 +73,9 @@ struct regcache_descr
>> 
>>   /* Cached table containing the type of each register.  */
>>   struct type **register_type;
>> +
>> +  /* Size of the largest register.  */
>> +  long max_register_size;
> 
> Is this ever negative? Why not make it unsigned?

Should never be negative. Will change.

> 
>> };
>> 
>> static void *
>> @@ -125,7 +128,9 @@ init_regcache_descr (struct gdbarch *gdbarch)
>> 	descr->sizeof_register[i] = TYPE_LENGTH (descr->register_type[i]);
>> 	descr->register_offset[i] = offset;
>> 	offset += descr->sizeof_register[i];
>> -	gdb_assert (MAX_REGISTER_SIZE >= descr->sizeof_register[i]);
>> +        gdb_assert (MAX_REGISTER_SIZE >= descr->sizeof_register[i]);
>> +	descr->max_register_size = std::max (descr->max_register_size,
>> +                                         descr->sizeof_register[i]);
>>       }
>>     /* Set the real size of the raw register cache buffer.  */
>>     descr->sizeof_raw_registers = offset;
>> @@ -135,7 +140,9 @@ init_regcache_descr (struct gdbarch *gdbarch)
>> 	descr->sizeof_register[i] = TYPE_LENGTH (descr->register_type[i]);
>> 	descr->register_offset[i] = offset;
>> 	offset += descr->sizeof_register[i];
>> -	gdb_assert (MAX_REGISTER_SIZE >= descr->sizeof_register[i]);
>> +        gdb_assert (MAX_REGISTER_SIZE >= descr->sizeof_register[i]);
>> +        descr->max_register_size = std::max (descr->max_register_size,
>> +                                         descr->sizeof_register[i]);
>>       }
>>     /* Set the real size of the readonly register cache buffer.  */
>>     descr->sizeof_cooked_registers = offset;
>> @@ -187,6 +194,13 @@ regcache_register_size (const struct regcache *regcache, int n)
>>   return register_size (get_regcache_arch (regcache), n);
>> }
>> 
>> +long
> 
> Same as above, is this ever negative?

Ok.

> 
>> +max_register_size (struct gdbarch *gdbarch)
>> +{
>> +  struct regcache_descr *descr = regcache_descr (gdbarch);
> 
> Is descr ever NULL?

I don’t think so. Gdbarch.c ensures everything is initialised.
None of the other functions in this file ever check for null.

> 
>> +  return descr->max_register_size;
>> +}
>> +
>> /* The register cache for storing raw register values.  */
>> 
>> struct regcache
>> @@ -327,7 +341,7 @@ regcache_save (struct regcache *dst, regcache_cooked_read_ftype *cooked_read,
>> 	       void *src)
>> {
>>   struct gdbarch *gdbarch = dst->descr->gdbarch;
>> -  gdb_byte buf[MAX_REGISTER_SIZE];
>> +  gdb_byte *buf = (gdb_byte *) alloca (max_register_size (gdbarch));
>>   int regnum;
>> 
>>   /* The DST should be `read-only', if it wasn't then the save would
>> @@ -369,7 +383,7 @@ regcache_restore (struct regcache *dst,
>> 		  void *cooked_read_context)
>> {
>>   struct gdbarch *gdbarch = dst->descr->gdbarch;
>> -  gdb_byte buf[MAX_REGISTER_SIZE];
>> +  gdb_byte *buf = (gdb_byte *) alloca (max_register_size (gdbarch));
>>   int regnum;
>> 
>>   /* The dst had better not be read-only. If it is, the `restore'
>> @@ -1279,7 +1293,7 @@ regcache_dump (struct regcache *regcache, struct ui_file *file,
>>   int footnote_register_offset = 0;
>>   int footnote_register_type_name_null = 0;
>>   long register_offset = 0;
>> -  gdb_byte buf[MAX_REGISTER_SIZE];
>> +  gdb_byte *buf = (gdb_byte *) alloca (max_register_size (gdbarch));
>> 
>> #if 0
>>   fprintf_unfiltered (file, "nr_raw_registers %d\n",
>> diff --git a/gdb/remote.c b/gdb/remote.c
>> index 9247d43b094925ff397eb36b450eaba521adfc99..86856e6a6aba1967faaa8ef547f8a48fcc63c383 100644
>> --- a/gdb/remote.c
>> +++ b/gdb/remote.c
>> @@ -7752,9 +7752,10 @@ remote_fetch_registers (struct target_ops *ops,
>> static void
>> remote_prepare_to_store (struct target_ops *self, struct regcache *regcache)
>> {
>> +  struct gdbarch *gdbarch = get_regcache_arch (regcache);
>>   struct remote_arch_state *rsa = get_remote_arch_state ();
>>   int i;
>> -  gdb_byte buf[MAX_REGISTER_SIZE];
>> +  gdb_byte *buf = (gdb_byte *) alloca (max_register_size (gdbarch));
>> 
>>   /* Make sure the entire registers array is valid.  */
>>   switch (packet_support (PACKET_P))
>> 
>> 
>> 
> 
> My comment is the same as 2/3. Should we use a different data structure that can grow/shrink as one wishes?