Re: [PATCH 3/5] set/show code-cache

[Doug Evans <dje at google dot com>]

On Wed, Oct 23, 2013 at 1:27 AM, Yao Qi <yao@codesourcery.com> wrote:
> Similar to stack cache, in this patch, we add
> TARGET_OBJECT_CODE_MEMORY to read code from target and add a new
> option "set code-cache on|off" to control use code cache or not.
>
> The invalidation of code cache is identical to stack cache, with this
> patch, function target_dcache_invalidate invalidates both.  Command
> "set {stack,code}-cache" invalidates either of them.
>
> gdb:
>
> 2013-10-23  Yao Qi  <yao@codesourcery.com>
>
>         * target.c (struct target_dcache) <code>: New field.
>         (target_dcache_alloc): Initialize field 'code'.
>         (set_stack_cache_enabled_p): Invalidate corresponding dcache.
>         (code_cache_enabled_p_1): New.
>         (code_cache_enabled_p): New.
>         (set_code_cache_enabled_p): New function.
>         (show_code_cache_enabled_p): New function.
>         (target_dcache_xfree): Free code cache.
>         (target_dcache_invalidate): Invalidate code cache.
>         (memory_xfer_partial_1): Handle TARGET_OBJECT_CODE_MEMORY and
>         code_cache_enabled_p.
>         (target_xfer_partial): Likewise.
>         (target_read_code): New function.
>         (initialize_targets): Register command.
>         * target.h (enum target_object) <TARGET_OBJECT_CODE_MEMORY>: New.
>         (target_read_code): Declare.
> ---
>  gdb/target.c |   97 ++++++++++++++++++++++++++++++++++++++++++++++++++--------
>  gdb/target.h |    5 +++
>  2 files changed, 89 insertions(+), 13 deletions(-)
>
> diff --git a/gdb/target.c b/gdb/target.c
> index 2ae42a8..624d41a 100644
> --- a/gdb/target.c
> +++ b/gdb/target.c
> @@ -218,6 +218,9 @@ struct target_dcache
>    /* Cache the memory on stack and areas specified by memory
>       attributes.  */
>    DCACHE *general;
> +
> +  /* Cache the code.  */
> +  DCACHE *code;
>  };

While perhaps it usually won't be a problem (*1), it's odd to have two caches.
If I do x/10x $addr and then x/10i $addr will both caches get populated?
[Will there be latent bugs due to "storing the same thing twice"?
(which has been a source of bugs in gdb)  Plus there's the extra
complexity of keeping both in sync.]


---
(*1): Because it's rarer to request reading areas of memory containing
code vs disassembling it or reading prologues.