public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
From: Simon Marchi <simark@simark.ca>
To: John Baldwin <jhb@FreeBSD.org>, gdb-patches@sourceware.org
Subject: Re: [PATCH 1/5] regcache: Add collect/supply_regset variants that accept a register base.
Date: Tue, 22 Nov 2022 14:56:08 -0500	[thread overview]
Message-ID: <eebe3a5c-9675-fa8f-17a2-559d61e83ed1@simark.ca> (raw)
In-Reply-To: <20220708005816.9408-2-jhb@FreeBSD.org>

On 7/7/22 20:58, John Baldwin wrote:
> Some register sets described by an array of regcache_map_entry
> structures do not have fixed register numbers in their associated
> architecture but do describe a block of registers whose numbers are at
> fixed offsets relative to some base register value.  An example of
> this are the TLS register sets for the ARM and AArch64 architectures.
> 
> Currently OS-specific architectures create register maps and register
> sets dynamically using the register base number.  However, this
> requires duplicating the code to create the register map and register
> set.  To reduce duplication, add variants of the collect_regset and
> supply_regset regcache methods which accept a base register number.
> For valid register map entries (i.e. not REGCACHE_MAP_SKIP), add this
> base register number to the value from the map entry to determine the
> final register number.

The patch LGTM.  I have some small comments, once addressed you can put
my:

Approved-By: Simon Marchi <simon.marchi@efficios.com>


> ---
>  gdb/regcache.c | 28 +++++++++++++++++++++++++---
>  gdb/regcache.h | 24 ++++++++++++++++++++++--
>  2 files changed, 47 insertions(+), 5 deletions(-)
> 
> diff --git a/gdb/regcache.c b/gdb/regcache.c
> index 037659ef8fa..1db3d972ef8 100644
> --- a/gdb/regcache.c
> +++ b/gdb/regcache.c
> @@ -1180,7 +1180,7 @@ regcache::transfer_regset_register (struct regcache *out_regcache, int regnum,
>  /* See regcache.h.  */
>  
>  void
> -regcache::transfer_regset (const struct regset *regset,
> +regcache::transfer_regset (const struct regset *regset, int regbase,
>  			   struct regcache *out_regcache,
>  			   int regnum, const gdb_byte *in_buf,
>  			   gdb_byte *out_buf, size_t size) const
> @@ -1195,6 +1195,9 @@ regcache::transfer_regset (const struct regset *regset,
>        int regno = map->regno;
>        int slot_size = map->size;
>  
> +      if (regno != REGCACHE_MAP_SKIP)
> +	regno += regbase;
> +
>        if (slot_size == 0 && regno != REGCACHE_MAP_SKIP)
>  	slot_size = m_descr->sizeof_register[regno];
>  
> @@ -1242,7 +1245,18 @@ void
>  regcache::supply_regset (const struct regset *regset,
>  			 int regnum, const void *buf, size_t size)
>  {
> -  transfer_regset (regset, this, regnum, (const gdb_byte *) buf, nullptr, size);
> +  transfer_regset (regset, 0, this, regnum, (const gdb_byte *) buf, nullptr,
> +		   size);
> +}

Can the old regcache::{supply,collect}_regset (without regbase) become
trivial wrappers around the new ones (with regbase)?  I would put the
implementation directly in the .h if doing that.

> +
> +/* See regcache.h.  */
> +
> +void
> +regcache::supply_regset (const struct regset *regset, int regbase,
> +			 int regnum, const void *buf, size_t size)
> +{
> +  transfer_regset (regset, regbase, this, regnum, (const gdb_byte *) buf,
> +		   nullptr, size);
>  }
>  
>  /* Collect register REGNUM from REGCACHE to BUF, using the register
> @@ -1261,11 +1275,19 @@ void
>  regcache::collect_regset (const struct regset *regset,
>  			 int regnum, void *buf, size_t size) const
>  {
> -  transfer_regset (regset, nullptr, regnum, nullptr, (gdb_byte *) buf, size);
> +  transfer_regset (regset, 0, nullptr, regnum, nullptr, (gdb_byte *) buf, size);
>  }
>  
>  /* See regcache.h  */
>  
> +void
> +regcache::collect_regset (const struct regset *regset, int regbase,
> +			 int regnum, void *buf, size_t size) const
> +{
> +  transfer_regset (regset, regbase, nullptr, regnum, nullptr, (gdb_byte *) buf,
> +		   size);
> +}
> +
>  bool
>  regcache_map_supplies (const struct regcache_map_entry *map, int regnum,
>  		       struct gdbarch *gdbarch, size_t size)
> diff --git a/gdb/regcache.h b/gdb/regcache.h
> index 1dbba5ce9af..f01127b20fb 100644
> --- a/gdb/regcache.h
> +++ b/gdb/regcache.h
> @@ -150,6 +150,14 @@ extern void regcache_collect_regset (const struct regset *regset,
>  				     int regnum, void *buf, size_t size);
>  
>  
> +extern void regcache_supply_regset (const struct regset *regset, int regbase,
> +				    struct regcache *regcache,
> +				    int regnum, const void *buf,
> +				    size_t size);
> +extern void regcache_collect_regset (const struct regset *regset, int regbase,
> +				     const struct regcache *regcache,
> +				     int regnum, void *buf, size_t size);

These don't have a definition, I guess they are not needed.

(Could we make a patch that removes regcache_supply_regset and
regcache_collect_regset, since they just forward to the regcache
methods?)

> +     belonging to the regset, otherwise just the register numbered
> +     REGNUM.  The REGSET's 'regmap' field must point to an array of
> +     'struct regcache_map_entry'.  The valid register numbers in each
> +     entry in 'struct regcache_map_entry' are offset by REGBASE.  */
> +
> +  void supply_regset (const struct regset *regset, int regbase,
> +		      int regnum, const void *buf, size_t size);
> +
> +  void collect_regset (const struct regset *regset, int regbase, int regnum,
> +		       void *buf, size_t size) const;
> +
>    void supply_regset (const struct regset *regset,
>  		      int regnum, const void *buf, size_t size);
>  
> -
>    void collect_regset (const struct regset *regset, int regnum,
>  		       void *buf, size_t size) const;

Can you document the last two, just by saying something like "Same as
the above, but with REGBASE == 0"?

Simon

  reply	other threads:[~2022-11-22 19:56 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-07-08  0:58 [PATCH 0/5] Improve support for regcache_map_entry with variable " John Baldwin
2022-07-08  0:58 ` [PATCH 1/5] regcache: Add collect/supply_regset variants that accept a " John Baldwin
2022-11-22 19:56   ` Simon Marchi [this message]
2022-11-22 20:18     ` Simon Marchi
2022-11-22 22:32     ` John Baldwin
2022-07-08  0:58 ` [PATCH 2/5] fbsd-nat: Use regset supply/collect methods John Baldwin
2022-11-22 20:20   ` Simon Marchi
2022-07-08  0:58 ` [PATCH 3/5] fbsd-nat: Pass an optional register base to the register set helpers John Baldwin
2022-11-22 20:38   ` Simon Marchi
2022-11-22 22:16     ` John Baldwin
2022-11-22 22:25       ` John Baldwin
2022-07-08  0:58 ` [PATCH 4/5] arm-fbsd: Use a static regset for the TLS register set John Baldwin
2022-11-22 20:39   ` Simon Marchi
2022-07-08  0:58 ` [PATCH 5/5] aarch64-fbsd: " John Baldwin
2022-11-22 20:40   ` Simon Marchi
2022-07-21 14:53 ` [PATCH 0/5] Improve support for regcache_map_entry with variable register base John Baldwin
2022-08-22 18:11   ` [PING] " John Baldwin
2022-09-20 17:50     ` John Baldwin
2022-10-20 20:26       ` [PING 4] " John Baldwin
2022-11-21 18:21         ` [PING 5] " John Baldwin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=eebe3a5c-9675-fa8f-17a2-559d61e83ed1@simark.ca \
    --to=simark@simark.ca \
    --cc=gdb-patches@sourceware.org \
    --cc=jhb@FreeBSD.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).