[PATCH v3] gdb/csky support .reg2 for kernel 4.x and later

[PATCH v3] gdb/csky support .reg2 for kernel 4.x and later

[Jiangshuai Li via Gdb-patches]

When kernel's version >= 4.x, the size of .reg2 section will be 400.
Contents of .reg2 are {
    unsigned long vr[96];
    unsigned long fcr;
    unsigned long fesr;
    unsigned long fid;
    unsigned long reserved;
};

VR[96] means: (vr0~vr15) + (fr16~fr31), each Vector register is
128-bits, each Float register is 64 bits, the total size is
(4*96).

In addition, for fr0~fr15, each FRx is the lower 64 bits of the
corresponding VRx. So fr0~fr15 and vr0~vr15 regisetrs use the same
offset.
---
 gdb/csky-linux-tdep.c | 164 +++++++++++++++++++++++++++++++++++++-----
 gdb/csky-tdep.h       |   3 +
 2 files changed, 151 insertions(+), 16 deletions(-)

diff --git a/gdb/csky-linux-tdep.c b/gdb/csky-linux-tdep.c
index 440045e7713..08eb8f42768 100644
--- a/gdb/csky-linux-tdep.c
+++ b/gdb/csky-linux-tdep.c
@@ -36,6 +36,8 @@
 #define SIZEOF_CSKY_GREGSET 34*4
 /* Float regset fesr fsr fr0-fr31 for CK810.  */
 #define SIZEOF_CSKY_FREGSET 34*4
+/* Float regset vr0~vr15 fr15~fr31, reserved for CK810 when kernel 4.x.  */
+#define SIZEOF_CSKY_FREGSET_K4X  400
 
 /* Offset mapping table from core_section to regcache of general
    registers for ck810.  */
@@ -118,15 +120,83 @@ csky_supply_fregset (const struct regset *regset,
   int fregset_num = ARRAY_SIZE (csky_fregset_offset);
 
   gdb_assert (len >= SIZEOF_CSKY_FREGSET);
-  for (i = 0; i < fregset_num; i++)
+  if (len == SIZEOF_CSKY_FREGSET)
     {
-      if ((regnum == csky_fregset_offset[i] || regnum == -1)
-	  && csky_fregset_offset[i] != -1)
-	{
-	  int num = csky_fregset_offset[i];
-	  offset += register_size (gdbarch, num);
-	  regcache->raw_supply (csky_fregset_offset[i], fregs + offset);
-	}
+      for (i = 0; i < fregset_num; i++)
+        {
+          if ((regnum == csky_fregset_offset[i] || regnum == -1)
+              && csky_fregset_offset[i] != -1)
+            {
+              int num = csky_fregset_offset[i];
+              offset += register_size (gdbarch, num);
+              regcache->raw_supply (csky_fregset_offset[i], fregs + offset);
+            }
+        }
+    }
+  else if (len == SIZEOF_CSKY_FREGSET_K4X)
+    {
+      /* When kernel version >= 4.x, .reg2 size will be 400.
+         Contents is {
+           unsigned long vr[96];
+           unsigned long fcr;
+           unsigned long fesr;
+           unsigned long fid;
+           unsigned long reserved;
+         }
+         VR[96] means: (vr0~vr15) + (fr16~fr31), each Vector register is
+         128-bits, each Float register is 64 bits, the total size is
+         (4*96).
+
+         In addition, for fr0~fr15, each FRx is the lower 64 bits of the
+         corresponding VRx. So fr0~fr15 and vr0~vr15 regisetrs use the same
+         offset.  */
+      int fcr_regno[] = {122, 123, 121}; /* fcr, fesr, fid.  */
+
+      /* Supply vr0~vr15.  */
+      for (i = 0; i < 16; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, (CSKY_VR0_REGNUM + i)))
+            {
+              offset = 16 * i;
+              regcache->raw_supply (CSKY_VR0_REGNUM + i,
+                                    fregs + offset);
+            }
+        }
+      /* Supply fr0~fr15.  */
+      for (i = 0; i < 16; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, (CSKY_FR0_REGNUM + i)))
+            {
+              offset = 16 * i;
+              regcache->raw_supply (CSKY_FR0_REGNUM + i,
+                                    fregs + offset);
+            }
+        }
+      /* Supply fr16~fr31.  */
+      for (i = 0; i < 16; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, (CSKY_FR16_REGNUM + i)))
+            {
+              offset = (16 * 16) + (8 * i);
+              regcache->raw_supply (CSKY_FR16_REGNUM + i,
+                                    fregs + offset);
+            }
+        }
+     /* Supply fcr, fesr, fid.  */
+      for (i = 0; i < 3; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, fcr_regno[i]))
+            {
+              offset = (16 * 16) + (16 * 8) + (4 * i);
+              regcache->raw_supply (fcr_regno[i],
+                                    fregs + offset);
+            }
+        }
+    }
+  else
+    {
+      warning (_("Unknow size %ld of section .reg2, can not get value"
+                 " of float registers."), len);
     }
 }
 
@@ -144,14 +214,73 @@ csky_collect_fregset (const struct regset *regset,
   int offset = 0;
 
   gdb_assert (len >= SIZEOF_CSKY_FREGSET);
-  for (regno = 0; regno < fregset_num; regno++)
+  if (len == SIZEOF_CSKY_FREGSET)
+    {
+      for (regno = 0; regno < fregset_num; regno++)
+        {
+          if ((regnum == csky_fregset_offset[regno] || regnum == -1)
+               && csky_fregset_offset[regno] != -1)
+            {
+              offset += register_size (gdbarch, csky_fregset_offset[regno]);
+              regcache->raw_collect (regno, fregs + offset);
+            }
+        }
+    }
+  else if (len == SIZEOF_CSKY_FREGSET_K4X)
+    {
+      /* When kernel version >= 4.x, .reg2 size will be 400.
+         Contents is {
+           unsigned long vr[96];
+           unsigned long fcr;
+           unsigned long fesr;
+           unsigned long fid;
+           unsigned long reserved;
+         }
+         VR[96] means: (vr0~vr15) + (fr16~fr31), each Vector register is$
+         128-bits, each Float register is 64 bits, the total size is$
+         (4*96).$
+
+         In addition, for fr0~fr15, each FRx is the lower 64 bits of the$
+         corresponding VRx. So fr0~fr15 and vr0~vr15 regisetrs use the same$
+         offset.  */
+      int i = 0;
+      int fcr_regno[] = {122, 123, 121}; /* fcr, fesr, fid.  */
+
+      /* Supply vr0~vr15.  */
+      for (i = 0; i < 16; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, (CSKY_VR0_REGNUM + i)))
+            {
+              offset = 16 * i;
+              regcache ->raw_collect (CSKY_VR0_REGNUM + i,
+                                      fregs + offset);
+            }
+        }
+      /* Supply fr16~fr31.  */
+      for (i = 0; i < 16; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, (CSKY_FR16_REGNUM + i)))
+            {
+              offset = (16 * 16) + (8 * i);
+              regcache ->raw_collect (CSKY_FR16_REGNUM + i,
+                                      fregs + offset);
+            }
+        }
+      /* Supply fcr, fesr, fid.  */
+      for (i = 0; i < 3; i ++)
+        {
+          if (gdbarch_register_name (gdbarch, fcr_regno[i]))
+            {
+              offset = (16 * 16) + (16 * 8) + (4 * i);
+              regcache ->raw_collect (fcr_regno[i],
+                                      fregs + offset);
+            }
+        }
+    }
+  else
     {
-      if ((regnum == csky_fregset_offset[regno] || regnum == -1)
-	  && csky_fregset_offset[regno] != -1)
-	{
-	  offset += register_size (gdbarch, csky_fregset_offset[regno]);
-	  regcache->raw_collect (regno, fregs + offset);
-	}
+      warning (_("Unknow size %ld of section .reg2, will not set value"
+                 " of float registers."), len);
     }
 }
 
@@ -166,7 +295,10 @@ static const struct regset csky_regset_float =
 {
   NULL,
   csky_supply_fregset,
-  csky_collect_fregset
+  csky_collect_fregset,
+  /* Allow .reg2 to have a different size, and the size of .reg2 should
+     always be bigger than SIZEOF_CSKY_FREGSET.  */
+  1
 };
 
 /* Iterate over core file register note sections.  */
diff --git a/gdb/csky-tdep.h b/gdb/csky-tdep.h
index 4cfc0a5d086..3391a08c2e5 100644
--- a/gdb/csky-tdep.h
+++ b/gdb/csky-tdep.h
@@ -93,6 +93,9 @@ enum csky_regnum
   CSKY_PSR_REGNUM = CSKY_CR0_REGNUM,
 
   CSKY_MAX_REGISTER_SIZE = 16,
+
+  /* Actually, the max regs number should be 1187. But if the
+     gdb stub does not send a tdesc-xml file to gdb, 253 works. */
   CSKY_MAX_REGS = 253
 };
 
-- 
2.25.1

Re: [PATCH v3] gdb/csky support .reg2 for kernel 4.x and later

[Tom de Vries via Gdb-patches]

On 8/1/22 04:40, Jiangshuai Li via Gdb-patches wrote:
> +      warning (_("Unknow size %ld of section .reg2, can not get value"
> +                 " of float registers."), len);

This broke the build on the gdb-armhf-ubuntu20_04 builder ( see 
https://builder.sourceware.org/buildbot/#/builders/170/builds/246 ):
...
../../binutils-gdb/gdb/csky-linux-tdep.c: In function ‘void 
csky_supply_fregset(const regset*, regcache*, int, const void*, size_t)’:
../../binutils-gdb/gdb/csky-linux-tdep.c:194:18: error: format ‘%ld’ 
expects argument of type ‘long int’, but argument 2 has type ‘size_t’ 
{aka ‘unsigned int’} [-Werror=format=]
   194 |       warning (_("Unknow size %ld of section .reg2, can not get 
value"
       | 
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   195 |    " of float registers."), len);
...

There are uses of %z in gdb, but very few, so I suppose the way to fix 
this is to use %s and pulongest.

Thanks,
- Tom

Re: [PATCH v3] gdb/csky support .reg2 for kernel 4.x and later

[Tom de Vries via Gdb-patches]

On 8/9/22 04:29, jiangshuai_li wrote:
>  >On 8/1/22 04:40, Jiangshuai Li via Gdb-patches wrote:
>  >> +      warning (_("Unknow size %ld of section .reg2, can not get value"
>  >> +                 " of float registers."), len);
>  >
>  >This broke the build on the gdb-armhf-ubuntu20_04 builder ( see
>  >https://builder.sourceware.org/buildbot/#/builders/170/builds/246 ):
>  >...
>  >../../binutils-gdb/gdb/csky-linux-tdep.c: In function ‘void
>  >csky_supply_fregset(const regset*, regcache*, int, const void*, size_t)’:
>  >../../binutils-gdb/gdb/csky-linux-tdep.c:194:18: error: format ‘%ld’
>  >expects argument of type ‘long int’, but argument 2 has type ‘size_t’
>  >{aka ‘unsigned int’} [-Werror=format=]
>   >  194 |       warning (_("Unknow size %ld of section .reg2, can not get
>  >value"
>  >       |
>  >^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>  >   195 |    " of float registers."), len);
>  >...
>  >
>  >There are uses of %z in gdb, but very few, so I suppose the way to fix
>  >this is to use %s and pulongest.
> 
> Thanks, i have used your suggestion to fix it, a patch has been committed.
> 

Thanks for fixing it, and I see the gdb-armhf-ubuntu20_04 builder is 
back to green status, yay !

Thanks,
- Tom

[PATCH v3] gdb/csky support .reg2 for kernel 4.x and later

[jiangshuai_li via Gdb-patches]

>On 8/1/22 04:40, Jiangshuai Li via Gdb-patches wrote:
>> +      warning (_("Unknow size %ld of section .reg2, can not get value"
>> +                 " of float registers."), len);
>
>This broke the build on the gdb-armhf-ubuntu20_04 builder ( see 
>https://builder.sourceware.org/buildbot/#/builders/170/builds/246 ):
>...
>../../binutils-gdb/gdb/csky-linux-tdep.c: In function ‘void 
>csky_supply_fregset(const regset*, regcache*, int, const void*, size_t)’:
>../../binutils-gdb/gdb/csky-linux-tdep.c:194:18: error: format ‘%ld’ 
>expects argument of type ‘long int’, but argument 2 has type ‘size_t’ 
>{aka ‘unsigned int’} [-Werror=format=]
 >  194 |       warning (_("Unknow size %ld of section .reg2, can not get 
>value"
>       | 
>^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>   195 |    " of float registers."), len);
>...
>
>There are uses of %z in gdb, but very few, so I suppose the way to fix 
>this is to use %s and pulongest.

Thanks, i have used your suggestion to fix it, a patch has been committed.

Jiangshuai.