gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

[Jon Ringle]

[-- Attachment #1: Type: text/plain, Size: 69 bytes --]

Please see: http://sourceware.org/ml/gdb/2007-03/msg00099.html

Jon


[-- Attachment #2: gdbserver-ARCH_HAS_MMU.patch --]
[-- Type: text/plain, Size: 1915 bytes --]

Index: linux-low.c
===================================================================
RCS file: /cvs/src/src/gdb/gdbserver/linux-low.c,v
retrieving revision 1.53
diff -u -r1.53 linux-low.c
--- linux-low.c	26 Feb 2007 20:10:18 -0000	1.53
+++ linux-low.c	7 Mar 2007 04:14:22 -0000
@@ -42,6 +42,12 @@
 # define PTRACE_SETSIGINFO 0x4203
 #endif
 
+#ifdef __UCLIBC__
+#if !(defined(__UCLIBC_HAS_MMU__) || defined(__ARCH_HAS_MMU__))
+#define HAS_NOMMU
+#endif
+#endif
+
 /* ``all_threads'' is keyed by the LWP ID - it should be the thread ID instead,
    however.  This requires changing the ID in place when we go from !using_threads
    to using_threads, immediately.
@@ -143,7 +149,7 @@
   void *new_process;
   int pid;
 
-#if defined(__UCLIBC__) && !defined(__UCLIBC_HAS_MMU__)
+#if defined(__UCLIBC__) && defined(HAS_NOMMU)
   pid = vfork ();
 #else
   pid = fork ();
@@ -1594,7 +1600,7 @@
     return 0;
 }
 
-#if defined(__UCLIBC__) && !defined(__UCLIBC_HAS_MMU__)
+#if defined(__UCLIBC__) && defined(HAS_NOMMU)
 #if defined(__mcoldfire__)
 /* These should really be defined in the kernel's ptrace.h header.  */
 #define PT_TEXT_ADDR 49*4
@@ -1664,7 +1670,7 @@
   linux_remove_watchpoint,
   linux_stopped_by_watchpoint,
   linux_stopped_data_address,
-#if defined(__UCLIBC__) && !defined(__UCLIBC_HAS_MMU__)
+#if defined(__UCLIBC__) && defined(HAS_NOMMU)
   linux_read_offsets,
 #else
   NULL,
Index: server.c
===================================================================
RCS file: /cvs/src/src/gdb/gdbserver/server.c,v
retrieving revision 1.48
diff -u -r1.48 server.c
--- server.c	26 Feb 2007 20:10:18 -0000	1.48
+++ server.c	7 Mar 2007 04:14:23 -0000
@@ -255,7 +255,7 @@
 {
   static struct inferior_list_entry *thread_ptr;
 
-  if (strcmp ("qSymbol::", own_buf) == 0)
+  if (strcmp ("qSymbol:", own_buf) == 0)
     {
       if (the_target->look_up_symbols != NULL)
 	(*the_target->look_up_symbols) ();

Re: gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

[Daniel Jacobowitz]

On Tue, Mar 06, 2007 at 11:21:43PM -0500, Jon Ringle wrote:
> -  if (strcmp ("qSymbol::", own_buf) == 0)
> +  if (strcmp ("qSymbol:", own_buf) == 0)

What's this for?  It can't be right.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [SPAM] Re: gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

[Jon Ringle]

Daniel Jacobowitz wrote:
> On Tue, Mar 06, 2007 at 11:21:43PM -0500, Jon Ringle wrote:
>   
>> -  if (strcmp ("qSymbol::", own_buf) == 0)
>> +  if (strcmp ("qSymbol:", own_buf) == 0)
>>     
>
> What's this for?  It can't be right.
>
>   
Sorry, I must have fat-fingered a DEL key there at some point and didn't 
notice it when I posted the patch file.
Jon

Re: gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

[Michael Snyder]

On Tue, 2007-03-06 at 23:21 -0500, Jon Ringle wrote:
> Please see: http://sourceware.org/ml/gdb/2007-03/msg00099.html
> 
> Jon
> 
> plain text document attachment (gdbserver-ARCH_HAS_MMU.patch)
> Index: linux-low.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/gdbserver/linux-low.c,v
> retrieving revision 1.53
> diff -u -r1.53 linux-low.c
> --- linux-low.c	26 Feb 2007 20:10:18 -0000	1.53
> +++ linux-low.c	7 Mar 2007 04:14:22 -0000
> @@ -42,6 +42,12 @@
>  # define PTRACE_SETSIGINFO 0x4203
>  #endif
>  
> +#ifdef __UCLIBC__
> +#if !(defined(__UCLIBC_HAS_MMU__) || defined(__ARCH_HAS_MMU__))
> +#define HAS_NOMMU
> +#endif
> +#endif

Shouldn't it be !defined(__ARCH_HAS_MMU__)?


And please remove the change below -- I'm sure it has
nothing to do with the one above.

> Index: server.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/gdbserver/server.c,v
> retrieving revision 1.48
> diff -u -r1.48 server.c
> --- server.c	26 Feb 2007 20:10:18 -0000	1.48
> +++ server.c	7 Mar 2007 04:14:23 -0000
> @@ -255,7 +255,7 @@
>  {
>    static struct inferior_list_entry *thread_ptr;
>  
> -  if (strcmp ("qSymbol::", own_buf) == 0)
> +  if (strcmp ("qSymbol:", own_buf) == 0)
>      {
>        if (the_target->look_up_symbols != NULL)
>  	(*the_target->look_up_symbols) ();

Re: [SPAM] Re: gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

[Michael Snyder]

On Wed, 2007-03-07 at 08:12 -0500, Jon Ringle wrote:
> Daniel Jacobowitz wrote:
> > On Tue, Mar 06, 2007 at 11:21:43PM -0500, Jon Ringle wrote:
> >   
> >> -  if (strcmp ("qSymbol::", own_buf) == 0)
> >> +  if (strcmp ("qSymbol:", own_buf) == 0)
> >>     
> >
> > What's this for?  It can't be right.
> >
> >   
> Sorry, I must have fat-fingered a DEL key there at some point and didn't 
> notice it when I posted the patch file.

Well, by the way, if that change is in your working gdb, 
it's likely to prevent thread debugging from working, and
might interact with the code that looks at SIGUSR1...

Re: gdbserver: should use __ARCH_HAS_MMU__ rather than __UCLIBC_HAS_MMU__

[Daniel Jacobowitz]

On Tue, Mar 06, 2007 at 11:21:43PM -0500, Jon Ringle wrote:
> Please see: http://sourceware.org/ml/gdb/2007-03/msg00099.html

> Jon

I belatedly checked this in (the linux-low.c portion of it).  Thanks.


-- 
Daniel Jacobowitz
CodeSourcery