[Qemu-devel] [PATCH] Check if we really are in userspace

[Qemu-devel] [PATCH] Check if we really are in userspace

[Thiemo Seufer]

Hello All,

this tests for MIPS if we are in userspace when handling a TLB fault.
For some of the other architectures this may also be a bug, only
ppc and x86 handle this so far.


Thiemo


Index: cpu-exec.c
===================================================================
RCS file: /sources/qemu/qemu/cpu-exec.c,v
retrieving revision 1.73
diff -u -p -r1.73 cpu-exec.c
--- cpu-exec.c	8 Feb 2006 22:43:39 -0000	1.73
+++ cpu-exec.c	16 Feb 2006 15:21:03 -0000
@@ -1089,7 +1089,9 @@ static inline int handle_cpu_signal(unsi
     }
 
     /* see if it is an MMU fault */
-    ret = cpu_mips_handle_mmu_fault(env, address, is_write, 1, 0);
+    ret = cpu_mips_handle_mmu_fault(env, address, is_write,
+				    ((env->hflags & MIPS_HFLAG_MODE)
+				     == MIPS_HFLAG_UM), 0);
     if (ret < 0)
         return 0; /* not an MMU fault */
     if (ret == 0)

Re: [Qemu-devel] [PATCH] Check if we really are in userspace

[Paul Brook]

> Index: cpu-exec.c
> ===================================================================
> RCS file: /sources/qemu/qemu/cpu-exec.c,v
> retrieving revision 1.73
> diff -u -p -r1.73 cpu-exec.c
> --- cpu-exec.c	8 Feb 2006 22:43:39 -0000	1.73
> +++ cpu-exec.c	16 Feb 2006 15:21:03 -0000
> @@ -1089,7 +1089,9 @@ static inline int handle_cpu_signal(unsi
>      }
>
>      /* see if it is an MMU fault */
> -    ret = cpu_mips_handle_mmu_fault(env, address, is_write, 1, 0);
> +    ret = cpu_mips_handle_mmu_fault(env, address, is_write,
> +				    ((env->hflags & MIPS_HFLAG_MODE)
> +				     == MIPS_HFLAG_UM), 0);
>      if (ret < 0)
>          return 0; /* not an MMU fault */
>      if (ret == 0)

This code is only used when !defined(CONFIG_SOFTMMU), ie. usermode emulation.
The i386 code is a leftover of the now-dead qemu-fast. Presumably the ppc code 
was just copied from these.

Paul