[PATCH] S390: Fix crash when remote tdesc doesn't define vec128

[PATCH] S390: Fix crash when remote tdesc doesn't define vec128

[Andreas Arnez]

I've encountered a GDB crash when trying to read registers from a remote
stub that provided a target.xml with vector registers, but without the
'vec128' data type.  The crash is caused by NULL register type entries for
the "concatenated" pseudo-registers v0-v15.  These NULL entries are
introduced by the logic in s390_pseudo_register_type(), where the tdesc
type 'vec128' is returned unconditionally -- even if it doesn't exist (is
NULL).

The fixed logic for determining a "concatenated" vector register's type
now returns the type of the raw register v16 instead.  This also makes
sure that all vector register have the same type.
---
 gdb/s390-tdep.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/gdb/s390-tdep.c b/gdb/s390-tdep.c
index 81fa0329ea..23689aa71a 100644
--- a/gdb/s390-tdep.c
+++ b/gdb/s390-tdep.c
@@ -1275,8 +1275,9 @@ s390_pseudo_register_type (struct gdbarch *gdbarch, int regnum)
   if (regnum_is_gpr_full (tdep, regnum))
     return builtin_type (gdbarch)->builtin_uint64;
 
+  /* For the "concatenated" vector registers use the same type as v16.  */
   if (regnum_is_vxr_full (tdep, regnum))
-    return tdesc_find_type (gdbarch, "vec128");
+    return tdesc_register_type (gdbarch, S390_V16_REGNUM);
 
   internal_error (__FILE__, __LINE__, _("invalid regnum"));
 }
-- 
2.17.0

Re: [PATCH] S390: Fix crash when remote tdesc doesn't define vec128

[Kevin Buettner]

Hi Andreas,

It needs a ChangeLog entry, but aside from that it looks okay to me.

Kevin

On Thu, 18 Oct 2018 19:54:22 +0200
Andreas Arnez <arnez@linux.ibm.com> wrote:

> I've encountered a GDB crash when trying to read registers from a remote
> stub that provided a target.xml with vector registers, but without the
> 'vec128' data type.  The crash is caused by NULL register type entries for
> the "concatenated" pseudo-registers v0-v15.  These NULL entries are
> introduced by the logic in s390_pseudo_register_type(), where the tdesc
> type 'vec128' is returned unconditionally -- even if it doesn't exist (is
> NULL).
> 
> The fixed logic for determining a "concatenated" vector register's type
> now returns the type of the raw register v16 instead.  This also makes
> sure that all vector register have the same type.
> ---
>  gdb/s390-tdep.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/gdb/s390-tdep.c b/gdb/s390-tdep.c
> index 81fa0329ea..23689aa71a 100644
> --- a/gdb/s390-tdep.c
> +++ b/gdb/s390-tdep.c
> @@ -1275,8 +1275,9 @@ s390_pseudo_register_type (struct gdbarch *gdbarch, int regnum)
>    if (regnum_is_gpr_full (tdep, regnum))
>      return builtin_type (gdbarch)->builtin_uint64;
>  
> +  /* For the "concatenated" vector registers use the same type as v16.  */
>    if (regnum_is_vxr_full (tdep, regnum))
> -    return tdesc_find_type (gdbarch, "vec128");
> +    return tdesc_register_type (gdbarch, S390_V16_REGNUM);
>  
>    internal_error (__FILE__, __LINE__, _("invalid regnum"));
>  }
> -- 
> 2.17.0
> 

Re: [PATCH] S390: Fix crash when remote tdesc doesn't define vec128

[Andreas Arnez]

On Fri, Oct 19 2018, Kevin Buettner wrote:

> Hi Andreas,
>
> It needs a ChangeLog entry, but aside from that it looks okay to me.

Thanks.  Here's the missing ChangeLog entry:

gdb/ChangeLog:

	* s390-tdep.c (s390_pseudo_register_type): For v0-v15 don't yield
	the possibly non-existent tdesc type 'vec128', but the type of raw
	register v16 instead.

Pushed with that as commit #0667c50682.

--
Andreas