[v3] btf: Add support to BTF_KIND_ENUM64 type

  Hello,

The following is patch v3 to update BTF/CTF backend supporting
BTF_KIND_ENUM64 type. Changes from v2:

  + Add a new `dtd_enum_unsigned' field in `ctf_dtdef' to indicate
    signedness of the enum type.
  + Fix endianness for representing BTF enum 64-bits enumerators.
  + Add {little,big}-endian testcases.

Comments will be welcomed and appreciated!,

Kind regards,
guillermo

--

BTF supports 64-bits enumerators with following encoding:

  struct btf_type:
    name_off: 0 or offset to a valid C identifier
    info.kind_flag: 0 for unsigned, 1 for signed
    info.kind: BTF_KIND_ENUM64
    info.vlen: number of enum values
    size: 1/2/4/8

The btf_type is followed by info.vlen number of:

    struct btf_enum64
    {
      uint32_t name_off;   /* Offset in string section of enumerator name.  */
      uint32_t val_lo32;   /* lower 32-bit value for a 64-bit value Enumerator */
      uint32_t val_hi32;   /* high 32-bit value for a 64-bit value Enumerator */
    };

So, a new btf_enum64 structure was added to represent BTF_KIND_ENUM64
and a new field dtd_enum_unsigned in ctf_dtdef structure to distinguish
when CTF enum is a signed or unsigned type, later that information is
used to encode the BTF enum type.

gcc/ChangeLog:

	* btfout.cc (btf_calc_num_vbytes): Compute enumeration size depending of
	enumerator type btf_enum{,64}.
	(btf_asm_type): Update btf_kflag according to enumeration type sign
	using dtd_enum_unsigned field for both:  BTF_KIND_ENUM{,64}.
	(btf_asm_enum_const): New argument to represent the size of
	the BTF enum type, writing the enumerator constant value for
	32 bits, if it's 64 bits then explicitly writes lower 32-bits
	value and higher 32-bits value.
	(output_asm_btf_enum_list): Add enumeration size argument.
	* ctfc.cc (ctf_add_enum): New argument to represent CTF enum
	basic information.
	(ctf_add_generic): Use of ei_{name. size, unsigned} to build the
	dtd structure containing enumeration information.
	(ctf_add_enumerator): Update comment mention support for BTF
	enumeration in 64-bits.
	* ctfc.h (ctf_dmdef): Update dmd_value to HOST_WIDE_INT to allow
	use 32/64 bits enumerators.
	(ctf_enum_binfo): New type to represent CTF basic enum type
	information.
	(ctf_dtdef): New field to describe enum signedness.
	* dwarf2ctf.cc (gen_ctf_enumeration_type): Use of ctf_enum_binfo
	type to pass information to ctf_add_enum to build the enum type.

include/
	* btf.h (btf_enum64): Add new definition and new symbolic
	constant to BTF_KIND_ENUM64 and BTF_KF_ENUM_{UN,}SIGNED.

gcc/testsuite/ChangeLog:

	* gcc.dg/debug/btf/btf-enum-1.c: Update testcase, with correct
	info.kflags encoding.
	* gcc.dg/debug/btf/btf-enum64-be-1.c: New testcase.
	* gcc.dg/debug/btf/btf-enum64-le-1.c: New testcase.
---
 gcc/btfout.cc                                 | 30 ++++++++++---
 gcc/ctfc.cc                                   | 22 +++++-----
 gcc/ctfc.h                                    | 15 +++++--
 gcc/dwarf2ctf.cc                              |  8 +++-
 gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c   |  2 +-
 .../gcc.dg/debug/btf/btf-enum64-be-1.c        | 44 +++++++++++++++++++
 .../gcc.dg/debug/btf/btf-enum64-le-1.c        | 44 +++++++++++++++++++
 include/btf.h                                 | 19 ++++++--
 8 files changed, 160 insertions(+), 24 deletions(-)
 create mode 100644 gcc/testsuite/gcc.dg/debug/btf/btf-enum64-be-1.c
 create mode 100644 gcc/testsuite/gcc.dg/debug/btf/btf-enum64-le-1.c
  

Hi Guillermo,

On 10/14/22 8:55 PM, Guillermo E. Martinez wrote:
> Hello,
> 
> The following is patch v3 to update BTF/CTF backend supporting
> BTF_KIND_ENUM64 type. Changes from v2:
> 
>    + Add a new `dtd_enum_unsigned' field in `ctf_dtdef' to indicate
>      signedness of the enum type.
>    + Fix endianness for representing BTF enum 64-bits enumerators.
>    + Add {little,big}-endian testcases.
> 
> Comments will be welcomed and appreciated!,
> 
> Kind regards,
> guillermo
> 
> --
> 
> BTF supports 64-bits enumerators with following encoding:
> 
>    struct btf_type:
>      name_off: 0 or offset to a valid C identifier
>      info.kind_flag: 0 for unsigned, 1 for signed
>      info.kind: BTF_KIND_ENUM64
>      info.vlen: number of enum values
>      size: 1/2/4/8
> 
> The btf_type is followed by info.vlen number of:
> 
>      struct btf_enum64
>      {
>        uint32_t name_off;   /* Offset in string section of enumerator name.  */
>        uint32_t val_lo32;   /* lower 32-bit value for a 64-bit value Enumerator */
>        uint32_t val_hi32;   /* high 32-bit value for a 64-bit value Enumerator */
>      };
> 
> So, a new btf_enum64 structure was added to represent BTF_KIND_ENUM64
> and a new field dtd_enum_unsigned in ctf_dtdef structure to distinguish
> when CTF enum is a signed or unsigned type, later that information is
> used to encode the BTF enum type.
> 
> gcc/ChangeLog:
> 
> 	* btfout.cc (btf_calc_num_vbytes): Compute enumeration size depending of
> 	enumerator type btf_enum{,64}.
> 	(btf_asm_type): Update btf_kflag according to enumeration type sign
> 	using dtd_enum_unsigned field for both:  BTF_KIND_ENUM{,64}.
> 	(btf_asm_enum_const): New argument to represent the size of
> 	the BTF enum type, writing the enumerator constant value for
> 	32 bits, if it's 64 bits then explicitly writes lower 32-bits
> 	value and higher 32-bits value.
> 	(output_asm_btf_enum_list): Add enumeration size argument.
> 	* ctfc.cc (ctf_add_enum): New argument to represent CTF enum
> 	basic information.
> 	(ctf_add_generic): Use of ei_{name. size, unsigned} to build the
> 	dtd structure containing enumeration information.
> 	(ctf_add_enumerator): Update comment mention support for BTF
> 	enumeration in 64-bits.
> 	* ctfc.h (ctf_dmdef): Update dmd_value to HOST_WIDE_INT to allow
> 	use 32/64 bits enumerators.
> 	(ctf_enum_binfo): New type to represent CTF basic enum type
> 	information.
> 	(ctf_dtdef): New field to describe enum signedness.
> 	* dwarf2ctf.cc (gen_ctf_enumeration_type): Use of ctf_enum_binfo
> 	type to pass information to ctf_add_enum to build the enum type.
> 
> include/
> 	* btf.h (btf_enum64): Add new definition and new symbolic
> 	constant to BTF_KIND_ENUM64 and BTF_KF_ENUM_{UN,}SIGNED.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* gcc.dg/debug/btf/btf-enum-1.c: Update testcase, with correct
> 	info.kflags encoding.
> 	* gcc.dg/debug/btf/btf-enum64-be-1.c: New testcase.
> 	* gcc.dg/debug/btf/btf-enum64-le-1.c: New testcase.
> ---
>   gcc/btfout.cc                                 | 30 ++++++++++---
>   gcc/ctfc.cc                                   | 22 +++++-----
>   gcc/ctfc.h                                    | 15 +++++--
>   gcc/dwarf2ctf.cc                              |  8 +++-
>   gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c   |  2 +-
>   .../gcc.dg/debug/btf/btf-enum64-be-1.c        | 44 +++++++++++++++++++
>   .../gcc.dg/debug/btf/btf-enum64-le-1.c        | 44 +++++++++++++++++++
>   include/btf.h                                 | 19 ++++++--
>   8 files changed, 160 insertions(+), 24 deletions(-)
>   create mode 100644 gcc/testsuite/gcc.dg/debug/btf/btf-enum64-be-1.c
>   create mode 100644 gcc/testsuite/gcc.dg/debug/btf/btf-enum64-le-1.c
> 
> diff --git a/gcc/btfout.cc b/gcc/btfout.cc
> index 997a33fa089..aef9fd70a28 100644
> --- a/gcc/btfout.cc
> +++ b/gcc/btfout.cc
> @@ -223,7 +223,9 @@ btf_calc_num_vbytes (ctf_dtdef_ref dtd)
>         break;
>   
>       case BTF_KIND_ENUM:
> -      vlen_bytes += vlen * sizeof (struct btf_enum);
> +      vlen_bytes += (dtd->dtd_data.ctti_size == 0x8)
> +			? vlen * sizeof (struct btf_enum64)
> +			: vlen * sizeof (struct btf_enum);
>         break;
>   
>       case BTF_KIND_FUNC_PROTO:
> @@ -622,6 +624,15 @@ btf_asm_type (ctf_container_ref ctfc, ctf_dtdef_ref dtd)
>         btf_size_type = 0;
>       }
>   
> +  if (btf_kind == BTF_KIND_ENUM)
> +    {
> +      btf_kflag = dtd->dtd_enum_unsigned
> +		    ? BTF_KF_ENUM_UNSIGNED
> +		    : BTF_KF_ENUM_SIGNED;
> +      if (dtd->dtd_data.ctti_size == 0x8)
> +	btf_kind = BTF_KIND_ENUM64;
> +   }
> +
>     dw2_asm_output_data (4, dtd->dtd_data.ctti_name, "btt_name");
>     dw2_asm_output_data (4, BTF_TYPE_INFO (btf_kind, btf_kflag, btf_vlen),
>   		       "btt_info: kind=%u, kflag=%u, vlen=%u",
> @@ -634,6 +645,7 @@ btf_asm_type (ctf_container_ref ctfc, ctf_dtdef_ref dtd)
>       case BTF_KIND_UNION:
>       case BTF_KIND_ENUM:
>       case BTF_KIND_DATASEC:
> +    case BTF_KIND_ENUM64:
>         dw2_asm_output_data (4, dtd->dtd_data.ctti_size, "btt_size: %uB",
>   			   dtd->dtd_data.ctti_size);
>         return;
> @@ -707,13 +719,19 @@ btf_asm_sou_member (ctf_container_ref ctfc, ctf_dmdef_t * dmd)
>       }
>   }
>   
> -/* Asm'out an enum constant following a BTF_KIND_ENUM.  */
> +/* Asm'out an enum constant following a BTF_KIND_ENUM{,64}.  */
>   
>   static void
> -btf_asm_enum_const (ctf_dmdef_t * dmd)
> +btf_asm_enum_const (unsigned int size, ctf_dmdef_t * dmd)
>   {
>     dw2_asm_output_data (4, dmd->dmd_name_offset, "bte_name");
> -  dw2_asm_output_data (4, dmd->dmd_value, "bte_value");
> +  if (size == 4)
> +    dw2_asm_output_data (size, dmd->dmd_value, "bte_value");
> +  else
> +    {
> +      dw2_asm_output_data (4, dmd->dmd_value & 0xffffffff, "bte_value_lo32");
> +      dw2_asm_output_data (4, (dmd->dmd_value >> 32) & 0xffffffff, "bte_value_hi32");
> +    }
>   }
>   
>   /* Asm'out a function parameter description following a BTF_KIND_FUNC_PROTO.  */
> @@ -871,7 +889,7 @@ output_asm_btf_sou_fields (ctf_container_ref ctfc, ctf_dtdef_ref dtd)
>         btf_asm_sou_member (ctfc, dmd);
>   }
>   
> -/* Output all enumerator constants following a BTF_KIND_ENUM.  */
> +/* Output all enumerator constants following a BTF_KIND_ENUM{,64}.  */
>   
>   static void
>   output_asm_btf_enum_list (ctf_container_ref ARG_UNUSED (ctfc),
> @@ -881,7 +899,7 @@ output_asm_btf_enum_list (ctf_container_ref ARG_UNUSED (ctfc),
>   
>     for (dmd = dtd->dtd_u.dtu_members;
>          dmd != NULL; dmd = (ctf_dmdef_t *) ctf_dmd_list_next (dmd))
> -    btf_asm_enum_const (dmd);
> +    btf_asm_enum_const (dtd->dtd_data.ctti_size, dmd);
>   }
>   
>   /* Output all function arguments following a BTF_KIND_FUNC_PROTO.  */
> diff --git a/gcc/ctfc.cc b/gcc/ctfc.cc
> index 9773358a475..80322eea91a 100644
> --- a/gcc/ctfc.cc
> +++ b/gcc/ctfc.cc
> @@ -576,8 +576,8 @@ ctf_add_array (ctf_container_ref ctfc, uint32_t flag, const ctf_arinfo_t * arp,
>   }
>   
>   ctf_id_t
> -ctf_add_enum (ctf_container_ref ctfc, uint32_t flag, const char * name,
> -	      HOST_WIDE_INT size, dw_die_ref die)
> +ctf_add_enum (ctf_container_ref ctfc, uint32_t flag,
> +	      const ctf_enum_binfo_t *ei, dw_die_ref die)

The name and size information is typically being passed via explicit 
arguments in the other similar APIs in this functionality.  I have a 
slight preference towards keeping it that way when possible.  So in this 
API ctf_add_enum, how about just adding another function argument for 
signedness and getting rid of the data structure ctf_enum_binfo_t 
altogether. What do you think ?

Patch looks good to me otherwise.

Thanks
Indu

>   {
>     ctf_dtdef_ref dtd;
>     ctf_id_t type;
> @@ -595,16 +595,16 @@ ctf_add_enum (ctf_container_ref ctfc, uint32_t flag, const char * name,
>   	= CTF_TYPE_INFO (CTF_K_FORWARD, CTF_ADD_NONROOT, 0);
>       }
>   
> -  type = ctf_add_generic (ctfc, flag, name, &dtd, die);
> +  type = ctf_add_generic (ctfc, flag, ei->ei_name, &dtd, die);
>   
>     dtd->dtd_data.ctti_info = CTF_TYPE_INFO (CTF_K_ENUM, flag, 0);
>   
>     /* Size in bytes should always fit, of course.
>        TBD WARN - warn instead?  */
> -  gcc_assert (size <= CTF_MAX_SIZE);
> -
> -  dtd->dtd_data.ctti_size = size;
> +  gcc_assert (ei->ei_size <= CTF_MAX_SIZE);
>   
> +  dtd->dtd_data.ctti_size = ei->ei_size;
> +  dtd->dtd_enum_unsigned = ei->ei_unsigned;
>     ctfc->ctfc_num_stypes++;
>   
>     return type;
> @@ -630,10 +630,12 @@ ctf_add_enumerator (ctf_container_ref ctfc, ctf_id_t enid, const char * name,
>   
>     gcc_assert (kind == CTF_K_ENUM && vlen < CTF_MAX_VLEN);
>   
> -  /* Enum value is of type HOST_WIDE_INT in the compiler, dmd_value is int32_t
> -     on the other hand.  Check bounds and skip adding this enum value if out of
> -     bounds.  */
> -  if ((value > INT_MAX) || (value < INT_MIN))
> +  /* Enum value is of type HOST_WIDE_INT in the compiler, CTF enumerators
> +     values in ctf_enum_t is limited to int32_t, BTF supports signed and
> +     unsigned enumerators values of 32 and 64 bits, for both debug formats
> +     we use ctf_dmdef_t.dmd_value entry of HOST_WIDE_INT type. So check
> +     CTF bounds and skip adding this enum value if out of bounds.  */
> +  if (!btf_debuginfo_p() && ((value > INT_MAX) || (value < INT_MIN)))
>       {
>         /* FIXME - Note this TBD_CTF_REPRESENTATION_LIMIT.  */
>         return (1);
> diff --git a/gcc/ctfc.h b/gcc/ctfc.h
> index bcf3a43ae1b..d3f50454385 100644
> --- a/gcc/ctfc.h
> +++ b/gcc/ctfc.h
> @@ -125,6 +125,14 @@ typedef struct GTY (()) ctf_itype
>   
>   #define CTF_FUNC_VARARG 0x1
>   
> +/* Basic enum information to build ctf_dmdef_t type.  */
> +typedef struct ctf_enum_binfo
> +{
> +  const char * ei_name;
> +  unsigned int ei_size;	    /* Size in bytes.  */
> +  bool ei_unsigned;
> +} ctf_enum_binfo_t;
> +
>   /* Struct/union/enum member definition for CTF generation.  */
>   
>   typedef struct GTY ((chain_next ("%h.dmd_next"))) ctf_dmdef
> @@ -133,7 +141,7 @@ typedef struct GTY ((chain_next ("%h.dmd_next"))) ctf_dmdef
>     ctf_id_t dmd_type;		/* Type of this member (for sou).  */
>     uint32_t dmd_name_offset;	/* Offset of the name in str table.  */
>     uint64_t dmd_offset;		/* Offset of this member in bits (for sou).  */
> -  int dmd_value;		/* Value of this member (for enum).  */
> +  HOST_WIDE_INT dmd_value;	/* Value of this member (for enum).  */
>     struct ctf_dmdef * dmd_next;	/* A list node.  */
>   } ctf_dmdef_t;
>   
> @@ -162,6 +170,7 @@ struct GTY ((for_user)) ctf_dtdef
>     bool from_global_func; /* Whether this type was added from a global
>   			    function.  */
>     uint32_t linkage;           /* Used in function types.  0=local, 1=global.  */
> +  bool dtd_enum_unsigned;     /* Enum signedness.  */
>     union GTY ((desc ("ctf_dtu_d_union_selector (&%1)")))
>     {
>       /* struct, union, or enum.  */
> @@ -405,8 +414,8 @@ extern const char * ctf_add_string (ctf_container_ref, const char *,
>   
>   extern ctf_id_t ctf_add_reftype (ctf_container_ref, uint32_t, ctf_id_t,
>   				 uint32_t, dw_die_ref);
> -extern ctf_id_t ctf_add_enum (ctf_container_ref, uint32_t, const char *,
> -			      HOST_WIDE_INT, dw_die_ref);
> +extern ctf_id_t ctf_add_enum (ctf_container_ref, uint32_t,
> +			      const ctf_enum_binfo_t *, dw_die_ref);
>   extern ctf_id_t ctf_add_slice (ctf_container_ref, uint32_t, ctf_id_t,
>   			       uint32_t, uint32_t, dw_die_ref);
>   extern ctf_id_t ctf_add_float (ctf_container_ref, uint32_t, const char *,
> diff --git a/gcc/dwarf2ctf.cc b/gcc/dwarf2ctf.cc
> index 397100004c2..56dfcadd3fb 100644
> --- a/gcc/dwarf2ctf.cc
> +++ b/gcc/dwarf2ctf.cc
> @@ -736,6 +736,7 @@ gen_ctf_enumeration_type (ctf_container_ref ctfc, dw_die_ref enumeration)
>   {
>     const char *enum_name = get_AT_string (enumeration, DW_AT_name);
>     unsigned int bit_size = ctf_die_bitsize (enumeration);
> +  unsigned int signedness = get_AT_unsigned (enumeration, DW_AT_encoding);
>     int declaration_p = get_AT_flag (enumeration, DW_AT_declaration);
>   
>     ctf_id_t enumeration_type_id;
> @@ -757,9 +758,14 @@ gen_ctf_enumeration_type (ctf_container_ref ctfc, dw_die_ref enumeration)
>         bit_size = ctf_die_bitsize (type);
>       }
>   
> +  ctf_enum_binfo_t ei;
> +  ei.ei_name = enum_name;
> +  ei.ei_size = bit_size / 8;
> +  ei.ei_unsigned = (signedness == DW_ATE_unsigned);
> +
>     /* Generate a CTF type for the enumeration.  */
>     enumeration_type_id = ctf_add_enum (ctfc, CTF_ADD_ROOT,
> -				      enum_name, bit_size / 8, enumeration);
> +				      &ei, enumeration);
>   
>     /* Process the enumerators.  */
>     {
> diff --git a/gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c b/gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c
> index 728493b0804..7e940529f1b 100644
> --- a/gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c
> +++ b/gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c
> @@ -4,7 +4,7 @@
>   /* { dg-options "-O0 -gbtf -dA" } */
>   
>   /* { dg-final { scan-assembler-times "\[\t \]0x6000004\[\t \]+\[^\n\]*btt_info" 1 } } */
> -/* { dg-final { scan-assembler-times "\[\t \]0x6000003\[\t \]+\[^\n\]*btt_info" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0x86000003\[\t \]+\[^\n\]*btt_info" 1 } } */
>   /* { dg-final { scan-assembler-times "ascii \"QAD.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
>   /* { dg-final { scan-assembler-times "ascii \"QED.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
>   /* { dg-final { scan-assembler-times "ascii \"QOD.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> diff --git a/gcc/testsuite/gcc.dg/debug/btf/btf-enum64-be-1.c b/gcc/testsuite/gcc.dg/debug/btf/btf-enum64-be-1.c
> new file mode 100644
> index 00000000000..40d44556e16
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/debug/btf/btf-enum64-be-1.c
> @@ -0,0 +1,44 @@
> +/* Test BTF generation for 64 bits enums.  */
> +
> +/* { dg-do compile } */
> +/* { dg-options "-O0 -gbtf -dA -mbig-endian" } */
> +
> +/* { dg-final { scan-assembler-times "\[\t \].size\[\t \]myenum1,\[\t \]8" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \].size\[\t \]myenum2,\[\t \]8" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \].size\[\t \]myenum3,\[\t \]8" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0x13000003\[\t \]+\[^\n\]*btt_info" 2 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0x93000003\[\t \]+\[^\n\]*btt_info" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0xffffffaa\[\t \]+\[^\n\]*bte_value_lo32" 2 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0xff\[\t \]+\[^\n\]*bte_value_hi32" 3 } } */
> +/* { dg-final { scan-assembler-times "ascii \"B1.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"B2.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"B3.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"C1.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"C2.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"C3.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"D1.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"D2.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"D3.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "bte_value_lo32" 9 } } */
> +/* { dg-final { scan-assembler-times "bte_value_hi32" 9 } } */
> +
> +enum default_enum
> +{
> +  B1 = 0xffffffffaa,
> +  B2 = 0xbbbbbbbb,
> +  B3 = 0xaabbccdd,
> +} myenum1 = B1;
> +
> +enum explicit_unsigned
> +{
> +  C1 = 0xffffffffbbUL,
> +  C2 = 0xbbbbbbbb,
> +  C3 = 0xaabbccdd,
> +} myenum2 = C1;
> +
> +enum signed64
> +{
> +  D1 = 0xffffffffaa,
> +  D2 = 0xbbbbbbbb,
> +  D3 = -0x1,
> +} myenum3 = D1;
> diff --git a/gcc/testsuite/gcc.dg/debug/btf/btf-enum64-le-1.c b/gcc/testsuite/gcc.dg/debug/btf/btf-enum64-le-1.c
> new file mode 100644
> index 00000000000..873a50ea174
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/debug/btf/btf-enum64-le-1.c
> @@ -0,0 +1,44 @@
> +/* Test BTF generation for 64 bits enums.  */
> +
> +/* { dg-do compile } */
> +/* { dg-options "-O0 -gbtf -dA -mlittle-endian" } */
> +
> +/* { dg-final { scan-assembler-times "\[\t \].size\[\t \]myenum1,\[\t \]8" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \].size\[\t \]myenum2,\[\t \]8" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \].size\[\t \]myenum3,\[\t \]8" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0x13000003\[\t \]+\[^\n\]*btt_info" 2 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0x93000003\[\t \]+\[^\n\]*btt_info" 1 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0xffffffaa\[\t \]+\[^\n\]*bte_value_lo32" 2 } } */
> +/* { dg-final { scan-assembler-times "\[\t \]0xff\[\t \]+\[^\n\]*bte_value_hi32" 3 } } */
> +/* { dg-final { scan-assembler-times "ascii \"B1.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"B2.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"B3.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"C1.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"C2.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"C3.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"D1.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"D2.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "ascii \"D3.0\"\[\t \]+\[^\n\]*btf_string" 1 } } */
> +/* { dg-final { scan-assembler-times "bte_value_lo32" 9 } } */
> +/* { dg-final { scan-assembler-times "bte_value_hi32" 9 } } */
> +
> +enum default_enum
> +{
> +  B1 = 0xffffffffaa,
> +  B2 = 0xbbbbbbbb,
> +  B3 = 0xaabbccdd,
> +} myenum1 = B1;
> +
> +enum explicit_unsigned
> +{
> +  C1 = 0xffffffffbbUL,
> +  C2 = 0xbbbbbbbb,
> +  C3 = 0xaabbccdd,
> +} myenum2 = C1;
> +
> +enum signed64
> +{
> +  D1 = 0xffffffffaa,
> +  D2 = 0xbbbbbbbb,
> +  D3 = -0x1,
> +} myenum3 = D1;
> diff --git a/include/btf.h b/include/btf.h
> index 78b551ced23..eba67f9d599 100644
> --- a/include/btf.h
> +++ b/include/btf.h
> @@ -109,7 +109,8 @@ struct btf_type
>   #define BTF_KIND_VAR		14	/* Variable.  */
>   #define BTF_KIND_DATASEC	15	/* Section such as .bss or .data.  */
>   #define BTF_KIND_FLOAT		16	/* Floating point.  */
> -#define BTF_KIND_MAX		BTF_KIND_FLOAT
> +#define BTF_KIND_ENUM64 	19	/* Enumeration up to 64 bits.  */
> +#define BTF_KIND_MAX		BTF_KIND_ENUM64
>   #define NR_BTF_KINDS		(BTF_KIND_MAX + 1)
>   
>   /* For some BTF_KINDs, struct btf_type is immediately followed by
> @@ -130,14 +131,17 @@ struct btf_type
>   #define BTF_INT_BOOL	(1 << 2)
>   
>   /* BTF_KIND_ENUM is followed by VLEN struct btf_enum entries,
> -   which describe the enumerators. Note that BTF currently only
> -   supports signed 32-bit enumerator values.  */
> +   which describe the enumerators. */
>   struct btf_enum
>   {
>     uint32_t name_off;	/* Offset in string section of enumerator name.  */
>     int32_t  val;		/* Enumerator value.  */
>   };
>   
> +/* BTF_KF_ENUM_ holds the flags for kflags in BTF_KIND_ENUM{,64}.  */
> +#define BTF_KF_ENUM_UNSIGNED	(0)
> +#define BTF_KF_ENUM_SIGNED 	(1 << 0)
> +
>   /* BTF_KIND_ARRAY is followed by a single struct btf_array.  */
>   struct btf_array
>   {
> @@ -190,6 +194,15 @@ struct btf_var_secinfo
>     uint32_t size;	/* Size (in bytes) of variable.  */
>   };
>   
> +/* BTF_KIND_ENUM64 is followed by VLEN struct btf_enum64 entries,
> +   which describe the 64 bits enumerators.  */
> +struct btf_enum64
> +{
> +  uint32_t name_off;	/* Offset in string section of enumerator name.  */
> +  uint32_t val_lo32;	/* lower 32-bit value for a 64-bit value Enumerator */
> +  uint32_t val_hi32;	/* high 32-bit value for a 64-bit value Enumerator */
> +};
> +
>   #ifdef	__cplusplus
>   }
>   #endif
>
  

On 10/18/22 16:40, Indu Bhagat wrote:
> Hi Guillermo,
> 

Hi Indu,

> On 10/14/22 8:55 PM, Guillermo E. Martinez wrote:
>> Hello,
>>
>> The following is patch v3 to update BTF/CTF backend supporting
>> BTF_KIND_ENUM64 type. Changes from v2:
>>
>>    + Add a new `dtd_enum_unsigned' field in `ctf_dtdef' to indicate
>>      signedness of the enum type.
>>    + Fix endianness for representing BTF enum 64-bits enumerators.
>>    + Add {little,big}-endian testcases.
>>
>> Comments will be welcomed and appreciated!,
>>
>> Kind regards,
>> guillermo
>>
>> -- 
>>
>> BTF supports 64-bits enumerators with following encoding:
>>
>>    struct btf_type:
>>      name_off: 0 or offset to a valid C identifier
>>      info.kind_flag: 0 for unsigned, 1 for signed
>>      info.kind: BTF_KIND_ENUM64
>>      info.vlen: number of enum values
>>      size: 1/2/4/8
>>
>> The btf_type is followed by info.vlen number of:
>>
>>      struct btf_enum64
>>      {
>>        uint32_t name_off;   /* Offset in string section of enumerator name.  */
>>        uint32_t val_lo32;   /* lower 32-bit value for a 64-bit value Enumerator */
>>        uint32_t val_hi32;   /* high 32-bit value for a 64-bit value Enumerator */
>>      };
>>
>> So, a new btf_enum64 structure was added to represent BTF_KIND_ENUM64
>> and a new field dtd_enum_unsigned in ctf_dtdef structure to distinguish
>> when CTF enum is a signed or unsigned type, later that information is
>> used to encode the BTF enum type.
>>
>> gcc/ChangeLog:
>>
>>     * btfout.cc (btf_calc_num_vbytes): Compute enumeration size depending of
>>     enumerator type btf_enum{,64}.
>>     (btf_asm_type): Update btf_kflag according to enumeration type sign
>>     using dtd_enum_unsigned field for both:  BTF_KIND_ENUM{,64}.
>>     (btf_asm_enum_const): New argument to represent the size of
>>     the BTF enum type, writing the enumerator constant value for
>>     32 bits, if it's 64 bits then explicitly writes lower 32-bits
>>     value and higher 32-bits value.
>>     (output_asm_btf_enum_list): Add enumeration size argument.
>>     * ctfc.cc (ctf_add_enum): New argument to represent CTF enum
>>     basic information.
>>     (ctf_add_generic): Use of ei_{name. size, unsigned} to build the
>>     dtd structure containing enumeration information.
>>     (ctf_add_enumerator): Update comment mention support for BTF
>>     enumeration in 64-bits.
>>     * ctfc.h (ctf_dmdef): Update dmd_value to HOST_WIDE_INT to allow
>>     use 32/64 bits enumerators.
>>     (ctf_enum_binfo): New type to represent CTF basic enum type
>>     information.
>>     (ctf_dtdef): New field to describe enum signedness.
>>     * dwarf2ctf.cc (gen_ctf_enumeration_type): Use of ctf_enum_binfo
>>     type to pass information to ctf_add_enum to build the enum type.
>>
>> include/
>>     * btf.h (btf_enum64): Add new definition and new symbolic
>>     constant to BTF_KIND_ENUM64 and BTF_KF_ENUM_{UN,}SIGNED.
>>
>> gcc/testsuite/ChangeLog:
>>
>>     * gcc.dg/debug/btf/btf-enum-1.c: Update testcase, with correct
>>     info.kflags encoding.
>>     * gcc.dg/debug/btf/btf-enum64-be-1.c: New testcase.
>>     * gcc.dg/debug/btf/btf-enum64-le-1.c: New testcase.
>> ---
>>   gcc/btfout.cc                                 | 30 ++++++++++---
>>   gcc/ctfc.cc                                   | 22 +++++-----
>>   gcc/ctfc.h                                    | 15 +++++--
>>   gcc/dwarf2ctf.cc                              |  8 +++-
>>   gcc/testsuite/gcc.dg/debug/btf/btf-enum-1.c   |  2 +-
>>   .../gcc.dg/debug/btf/btf-enum64-be-1.c        | 44 +++++++++++++++++++
>>   .../gcc.dg/debug/btf/btf-enum64-le-1.c        | 44 +++++++++++++++++++
>>   include/btf.h                                 | 19 ++++++--
>>   8 files changed, 160 insertions(+), 24 deletions(-)
>>   create mode 100644 gcc/testsuite/gcc.dg/debug/btf/btf-enum64-be-1.c
>>   create mode 100644 gcc/testsuite/gcc.dg/debug/btf/btf-enum64-le-1.c
>>
>> diff --git a/gcc/btfout.cc b/gcc/btfout.cc
>> index 997a33fa089..aef9fd70a28 100644
>> --- a/gcc/btfout.cc
>> +++ b/gcc/btfout.cc
>> @@ -223,7 +223,9 @@ btf_calc_num_vbytes (ctf_dtdef_ref dtd)
>>         break;
>>       case BTF_KIND_ENUM:
>> -      vlen_bytes += vlen * sizeof (struct btf_enum);
>> +      vlen_bytes += (dtd->dtd_data.ctti_size == 0x8)
>> +            ? vlen * sizeof (struct btf_enum64)
>> +            : vlen * sizeof (struct btf_enum);
>>         break;
>>       case BTF_KIND_FUNC_PROTO:
>> @@ -622,6 +624,15 @@ btf_asm_type (ctf_container_ref ctfc, ctf_dtdef_ref dtd)
>>         btf_size_type = 0;
>>       }
>> +  if (btf_kind == BTF_KIND_ENUM)
>> +    {
>> +      btf_kflag = dtd->dtd_enum_unsigned
>> +            ? BTF_KF_ENUM_UNSIGNED
>> +            : BTF_KF_ENUM_SIGNED;
>> +      if (dtd->dtd_data.ctti_size == 0x8)
>> +    btf_kind = BTF_KIND_ENUM64;
>> +   }
>> +
>>     dw2_asm_output_data (4, dtd->dtd_data.ctti_name, "btt_name");
>>     dw2_asm_output_data (4, BTF_TYPE_INFO (btf_kind, btf_kflag, btf_vlen),
>>                  "btt_info: kind=%u, kflag=%u, vlen=%u",
>> @@ -634,6 +645,7 @@ btf_asm_type (ctf_container_ref ctfc, ctf_dtdef_ref dtd)
>>       case BTF_KIND_UNION:
>>       case BTF_KIND_ENUM:
>>       case BTF_KIND_DATASEC:
>> +    case BTF_KIND_ENUM64:
>>         dw2_asm_output_data (4, dtd->dtd_data.ctti_size, "btt_size: %uB",
>>                  dtd->dtd_data.ctti_size);
>>         return;
>> @@ -707,13 +719,19 @@ btf_asm_sou_member (ctf_container_ref ctfc, ctf_dmdef_t * dmd)
>>       }
>>   }
>> -/* Asm'out an enum constant following a BTF_KIND_ENUM.  */
>> +/* Asm'out an enum constant following a BTF_KIND_ENUM{,64}.  */
>>   static void
>> -btf_asm_enum_const (ctf_dmdef_t * dmd)
>> +btf_asm_enum_const (unsigned int size, ctf_dmdef_t * dmd)
>>   {
>>     dw2_asm_output_data (4, dmd->dmd_name_offset, "bte_name");
>> -  dw2_asm_output_data (4, dmd->dmd_value, "bte_value");
>> +  if (size == 4)
>> +    dw2_asm_output_data (size, dmd->dmd_value, "bte_value");
>> +  else
>> +    {
>> +      dw2_asm_output_data (4, dmd->dmd_value & 0xffffffff, "bte_value_lo32");
>> +      dw2_asm_output_data (4, (dmd->dmd_value >> 32) & 0xffffffff, "bte_value_hi32");
>> +    }
>>   }
>>   /* Asm'out a function parameter description following a BTF_KIND_FUNC_PROTO.  */
>> @@ -871,7 +889,7 @@ output_asm_btf_sou_fields (ctf_container_ref ctfc, ctf_dtdef_ref dtd)
>>         btf_asm_sou_member (ctfc, dmd);
>>   }
>> -/* Output all enumerator constants following a BTF_KIND_ENUM.  */
>> +/* Output all enumerator constants following a BTF_KIND_ENUM{,64}.  */
>>   static void
>>   output_asm_btf_enum_list (ctf_container_ref ARG_UNUSED (ctfc),
>> @@ -881,7 +899,7 @@ output_asm_btf_enum_list (ctf_container_ref ARG_UNUSED (ctfc),
>>     for (dmd = dtd->dtd_u.dtu_members;
>>          dmd != NULL; dmd = (ctf_dmdef_t *) ctf_dmd_list_next (dmd))
>> -    btf_asm_enum_const (dmd);
>> +    btf_asm_enum_const (dtd->dtd_data.ctti_size, dmd);
>>   }
>>   /* Output all function arguments following a BTF_KIND_FUNC_PROTO.  */
>> diff --git a/gcc/ctfc.cc b/gcc/ctfc.cc
>> index 9773358a475..80322eea91a 100644
>> --- a/gcc/ctfc.cc
>> +++ b/gcc/ctfc.cc
>> @@ -576,8 +576,8 @@ ctf_add_array (ctf_container_ref ctfc, uint32_t flag, const ctf_arinfo_t * arp,
>>   }
>>   ctf_id_t
>> -ctf_add_enum (ctf_container_ref ctfc, uint32_t flag, const char * name,
>> -          HOST_WIDE_INT size, dw_die_ref die)
>> +ctf_add_enum (ctf_container_ref ctfc, uint32_t flag,
>> +          const ctf_enum_binfo_t *ei, dw_die_ref die)
> 
> The name and size information is typically being passed via explicit arguments in the other similar APIs in this functionality.  I have a slight preference towards keeping it that way when possible.  So in this API ctf_add_enum, how about just adding another function argument for signedness and getting rid of the data structure ctf_enum_binfo_t altogether. What do you think ?
> 

Sure!, I'll send the patch v4 with this change.

> Patch looks good to me otherwise.
>

Thanks for your comments,
  
> Thanks
> Indu
> 
>> [...]
Kind regards,
guillermo