Re: [RFA] Submit process record and replay fourth time, 4/8

[Pedro Alves <pedro@codesourcery.com>]

Hi Hui,

Let's just fix a few small issues, and then put this in.

(There are more issues here to solve here --- e.g., IIRC, Marc reported
success in using precord against gdbserver (that was only made possible
due to the fact that we managed to remove everything there used to 
be in linux-nat.c in previous versions and move those things to record.c),
but, it will only work when debugging from a x86-linux host,
cross debugging will not work correctly.  But, this is a minor thing
to handle at this point.)

On Wednesday 22 April 2009 10:06:42, Hui Zhu wrote:
> +                   if (record_debug)
> +                     fprintf_unfiltered (gdb_stdlog,
> +                                         "Process record: error reading "
> +                                         "memory at addr = 0x%s len = %d.\n",
> +                                         paddr_nz (tmpu32), sizeof (a));
> +                   return -1;


Printing `sizeof(foo)' with a %d formatter will fail to compile due
to a warning on 64-bit hosts, yet you've added linux-record.o to
the x86_64-*-linux* configuration.   Please fix that before checking in.
You have more instances of this in the patch.

On Wednesday 22 April 2009 10:06:42, Hui Zhu wrote:
> +   Copyright (C) 2008 Free Software Foundation, Inc.

2009.

+typedef struct linux_record_tdep_s
+{
...
+ } linux_record_tdep_t;

As in other parts of the series, this 'typedef struct foo_s foo_t'
doesn't fit GDB's style.  Just make this 'struct linux_record_tdep'
instead.

This patch is approved with the above fixes, and a changelog entry.

-- 
Pedro Alves