[obv] [s390] gdbserver regcache compilation fix

[obv] [s390] gdbserver regcache compilation fix

[Jan Kratochvil]

Hi Pedro,

checked-in this one as [obv] althought remains IMO-non-obv ppc:

linux-ppc-low.c: In function ‘ppc_arch_setup’:
linux-ppc-low.c:347: warning: passing argument 1 of ‘collect_register_by_name’ from incompatible pointer type
linux-ppc-low.c:347: warning: passing argument 2 of ‘collect_register_by_name’ from incompatible pointer type
linux-ppc-low.c:347: error: too few arguments to function ‘collect_register_by_name’


Thanks,
Jan


http://sourceware.org/ml/gdb-cvs/2010-01/msg00193.html

--- src/gdb/gdbserver/ChangeLog	2010/01/21 18:35:42	1.324
+++ src/gdb/gdbserver/ChangeLog	2010/01/21 19:26:49	1.325
@@ -1,3 +1,8 @@
+2010-01-21  Jan Kratochvil  <jan.kratochvil@redhat.com>
+
+	* linux-s390-low.c (s390_collect_ptrace_register)
+	(s390_supply_ptrace_register): Adjust it for the new regcache parameter.
+
 2010-01-21  Doug Evans  <dje@google.com>
 
 	* linux-low.c (PTRACE_ARG3_TYPE): Change from long to void*.
--- src/gdb/gdbserver/linux-s390-low.c	2010/01/20 22:55:38	1.19
+++ src/gdb/gdbserver/linux-s390-low.c	2010/01/21 19:26:49	1.20
@@ -118,8 +118,9 @@
       if ((regno ^ 1) < the_low_target.num_regs
 	  && the_low_target.regmap[regno ^ 1] == regaddr)
 	{
-	  collect_register (regno & ~1, buf);
-	  collect_register ((regno & ~1) + 1, buf + sizeof (long) - size);
+	  collect_register (regcache, regno & ~1, buf);
+	  collect_register (regcache, (regno & ~1) + 1,
+			    buf + sizeof (long) - size);
 	}
       else if (regaddr == PT_PSWADDR
 	       || (regaddr >= PT_GPR0 && regaddr <= PT_GPR15))
@@ -133,7 +134,7 @@
 	buf[size] |= 0x80;
     }
   else
-    collect_register (regno, buf);
+    collect_register (regcache, regno, buf);
 }
 
 static void
@@ -147,8 +148,9 @@
       if ((regno ^ 1) < the_low_target.num_regs
 	  && the_low_target.regmap[regno ^ 1] == regaddr)
 	{
-	  supply_register (regno & ~1, buf);
-	  supply_register ((regno & ~1) + 1, buf + sizeof (long) - size);
+	  supply_register (regcache, regno & ~1, buf);
+	  supply_register (regcache, (regno & ~1) + 1,
+			   buf + sizeof (long) - size);
 	}
       else if (regaddr == PT_PSWADDR
 	       || (regaddr >= PT_GPR0 && regaddr <= PT_GPR15))

Re: [obv] [s390] gdbserver regcache compilation fix

[Pedro Alves]

On Thursday 21 January 2010 19:28:52, Jan Kratochvil wrote:
> Hi Pedro,
> 
> checked-in this one as [obv] althought remains IMO-non-obv ppc:

Thanks.

> 
> linux-ppc-low.c: In function ‘ppc_arch_setup’:
> linux-ppc-low.c:347: warning: passing argument 1 of ‘collect_register_by_name’ from incompatible pointer type
> linux-ppc-low.c:347: warning: passing argument 2 of ‘collect_register_by_name’ from incompatible pointer type
> linux-ppc-low.c:347: error: too few arguments to function ‘collect_register_by_name’

Hmm, I see, I only built a 32-bit ppc gdbserver, and this is code
guarded on #ifdef __powerpc64__.

The fix becomes obvious considering that before my change,
collect_registers_by_... used to call get_thread_registers itself, the
change mostly just moved this call up to the caller sites.
I've applied this patch below, after confirming a 64-bit
ppc gdbserver now builds.  Let me know if something
still isn't working.

-- 
Pedro Alves

2010-01-21  Pedro Alves  <pedro@codesourcery.com>

	gdb/gdbserver/
	* linux-ppc-low.c (ppc_arch_setup): Adjust to regcache changes.

---
 gdb/gdbserver/linux-ppc-low.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: src/gdb/gdbserver/linux-ppc-low.c
===================================================================
--- src.orig/gdb/gdbserver/linux-ppc-low.c	2010-01-21 13:46:00.000000000 -0800
+++ src/gdb/gdbserver/linux-ppc-low.c	2010-01-21 13:46:09.000000000 -0800
@@ -335,6 +335,7 @@ ppc_arch_setup (void)
 {
 #ifdef __powerpc64__
   long msr;
+  struct regcache *regcache;
 
   /* On a 64-bit host, assume 64-bit inferior process with no
      AltiVec registers.  Reset ppc_hwcap to ensure that the
@@ -344,7 +345,8 @@ ppc_arch_setup (void)
 
   /* Only if the high bit of the MSR is set, we actually have
      a 64-bit inferior.  */
-  collect_register_by_name ("msr", &msr);
+  regcache = get_thread_regcache (current_inferior, 1);
+  collect_register_by_name (regcache, "msr", &msr);
   if (msr < 0)
     {
       ppc_get_hwcap (&ppc_hwcap);