This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH 1/3] Use register_size () instead of MAX_REGISTER_SIZE
- From: Alan Hayward <Alan dot Hayward at arm dot com>
- To: Luis Machado <lgustavo at codesourcery dot com>
- Cc: "gdb-patches at sourceware dot org" <gdb-patches at sourceware dot org>, nd <nd at arm dot com>
- Date: Tue, 10 Jan 2017 11:27:46 +0000
- Subject: Re: [PATCH 1/3] Use register_size () instead of MAX_REGISTER_SIZE
- Authentication-results: sourceware.org; auth=none
- Authentication-results: spf=none (sender IP is ) smtp.mailfrom=Alan dot Hayward at arm dot com;
- Nodisclaimer: True
- References: <902F747B-B822-4629-8DB8-D6775357E6A5@arm.com> <20085eca-52a3-ea33-9c9d-27f9064c986b@codesourcery.com>
- Spamdiagnosticmetadata: NSPM
- Spamdiagnosticoutput: 1:99
> On 9 Jan 2017, at 20:07, Luis Machado <lgustavo@codesourcery.com> wrote:
>
> On 01/09/2017 04:55 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 hold a single register, then
>> MAX_REGISTER_SIZE can be replaced with a call to register_size ().
>>
>> This patch is restricted to amd64-tdep.c, remote.c and regcache.c.
>> Follow on patch sets will expand to other files.
>>
>> Tested on x86.
>> Ok to commit?
>>
>> Thanks,
>> Alan.
>>
>> 2017-01-09 Alan Hayward <alan.hayward@arm.com>
>>
>> * amd64-tdep.c (amd64_pseudo_register_read_value): remove
>> MAX_REGISTER_SIZE.
>> (amd64_pseudo_register_read_value): Likewise.
>> * remote.c (fetch_register_using_p): Remove MAX_REGISTER_SIZE.
>> (store_register_using_P): Likewise.
>> * regcache.c (regcache_xfer_part): Likewise.
>>
>>
>> diff --git a/gdb/amd64-tdep.c b/gdb/amd64-tdep.c
>> index a6367424e75b234c89e05b2dba7490ac3e212c3c..fc1f134a18098862fab92ad45e55b05372027beb 100644
>> --- a/gdb/amd64-tdep.c
>> +++ b/gdb/amd64-tdep.c
>> @@ -353,7 +353,7 @@ amd64_pseudo_register_read_value (struct gdbarch *gdbarch,
>> struct regcache *regcache,
>> int regnum)
>> {
>> - gdb_byte raw_buf[MAX_REGISTER_SIZE];
>> + gdb_byte *raw_buf = (gdb_byte *) alloca (register_size (gdbarch, regnum));
>> struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
>> enum register_status status;
>> struct value *result_value;
>> @@ -414,7 +414,7 @@ amd64_pseudo_register_write (struct gdbarch *gdbarch,
>> struct regcache *regcache,
>> int regnum, const gdb_byte *buf)
>> {
>> - gdb_byte raw_buf[MAX_REGISTER_SIZE];
>> + gdb_byte *raw_buf = (gdb_byte *) alloca (register_size (gdbarch, regnum));
>> struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
>>
>> if (i386_byte_regnum_p (gdbarch, regnum))
>> diff --git a/gdb/regcache.c b/gdb/regcache.c
>> index b2b9524d742a477f89195cf657085833054cbb38..9d28aa2c2114e0f1c52758bb2fbe9669a329c13e 100644
>> --- a/gdb/regcache.c
>> +++ b/gdb/regcache.c
>> @@ -988,7 +988,8 @@ regcache_xfer_part (struct regcache *regcache, int regnum,
>> const gdb_byte *buf))
>> {
>> struct regcache_descr *descr = regcache->descr;
>> - gdb_byte reg[MAX_REGISTER_SIZE];
>> + struct gdbarch *gdbarch = get_regcache_arch (regcache);
>> + gdb_byte *reg = (gdb_byte *) alloca (register_size (gdbarch, regnum));
>>
>> gdb_assert (offset >= 0 && offset <= descr->sizeof_register[regnum]);
>> gdb_assert (len >= 0 && offset + len <= descr->sizeof_register[regnum]);
>> diff --git a/gdb/remote.c b/gdb/remote.c
>> index 837b9ee5124e8ac3f4e3a55279adae49c790be19..6da6eb366ae442354fd6a37741335af9a4a5a056 100644
>> --- a/gdb/remote.c
>> +++ b/gdb/remote.c
>> @@ -7462,9 +7462,10 @@ remote_wait (struct target_ops *ops,
>> static int
>> fetch_register_using_p (struct regcache *regcache, struct packet_reg *reg)
>> {
>> + struct gdbarch *gdbarch = get_regcache_arch (regcache);
>> struct remote_state *rs = get_remote_state ();
>> char *buf, *p;
>> - char regp[MAX_REGISTER_SIZE];
>> + gdb_byte *regp = (gdb_byte *) alloca (register_size (gdbarch, reg->regnum));
>> int i;
>>
>> if (packet_support (PACKET_p) == PACKET_DISABLE)
>> @@ -7769,7 +7770,7 @@ store_register_using_P (const struct regcache *regcache,
>> struct remote_state *rs = get_remote_state ();
>> /* Try storing a single register. */
>> char *buf = rs->buf;
>> - gdb_byte regp[MAX_REGISTER_SIZE];
>> + gdb_byte *regp = (gdb_byte *) alloca (register_size (gdbarch, reg->regnum));
>> char *p;
>>
>> if (packet_support (PACKET_P) == PACKET_DISABLE)
>>
>>
>>
>>
>
> Patch looks OK to me. Are you planning on getting rid of all/most of the MAX_REGISTER_SIZE in gdb?
Yes, I plan to remove the define completely, because Aarch64 SVE support would have required
MAX_REGISTER_SIZE to be changed from 64 to 256.
Thanks for the review.
Alan.