https://gcc.gnu.org/g:50efa788a4b6c7f619101b6156b47b91124c6dd3
commit r16-3723-g50efa788a4b6c7f619101b6156b47b91124c6dd3 Author: David Faust <[email protected]> Date: Thu Aug 28 09:12:55 2025 -0700 ctf: fix integer truncations in very large structs [PR121411] DWARF to CTF translation for type bit sizes was using uint32_t, and for member offsets was inadvertently using unsigned int via get_AT_unsigned. For very large struct types, at least one of these could be truncated causing incorrect encoding of the struct type and member offsets. Use HOST_WIDE_INT to avoid these truncation issues and fix the encoding for large structs. PR debug/121411 gcc/ * dwarf2ctf.cc (ctf_get_AT_data_member_location) Use AT_unsigned when fetching AT_bit_offset and AT_data_member_location. Simplify. (ctf_die_bitsize): Return unsigned HOST_WIDE_INT instead of uint32_t. (gen_ctf_base_type, gen_ctf_sou_type, gen_ctf_enumeration_type): Adapt accordingly. gcc/testsuite/ * gcc.dg/debug/ctf/ctf-struct-3.c: New test. Diff: --- gcc/dwarf2ctf.cc | 34 +++++++++++++-------------- gcc/testsuite/gcc.dg/debug/ctf/ctf-struct-3.c | 32 +++++++++++++++++++++++++ 2 files changed, 48 insertions(+), 18 deletions(-) diff --git a/gcc/dwarf2ctf.cc b/gcc/dwarf2ctf.cc index f8b305b6c4a8..4b49b23f0780 100644 --- a/gcc/dwarf2ctf.cc +++ b/gcc/dwarf2ctf.cc @@ -87,8 +87,8 @@ ctf_get_AT_data_member_location (dw_die_ref die) /* The field location (in bits) can be determined from either a DW_AT_data_member_location attribute or a DW_AT_data_bit_offset attribute. */ - if (get_AT (die, DW_AT_data_bit_offset)) - field_location = get_AT_unsigned (die, DW_AT_data_bit_offset); + if ((attr = get_AT (die, DW_AT_data_bit_offset))) + field_location = AT_unsigned (attr); else { attr = get_AT (die, DW_AT_data_member_location); @@ -102,16 +102,12 @@ ctf_get_AT_data_member_location (dw_die_ref die) == dw_val_class_unsigned_const); field_location = descr->dw_loc_oprnd1.v.val_unsigned * 8; } - else - { - attr = get_AT (die, DW_AT_data_member_location); - if (attr && AT_class (attr) == dw_val_class_const) - field_location = AT_int (attr) * 8; - else - field_location = (get_AT_unsigned (die, - DW_AT_data_member_location) - * 8); - } + else if (attr && AT_class (attr) == dw_val_class_const) + field_location = AT_int (attr) * 8; + else if (attr) + field_location = AT_unsigned (attr) * 8; + + /* Otherwise the location is non-existant, e.g. for union members. */ } return field_location; @@ -199,7 +195,7 @@ gen_ctf_unknown_type (ctf_container_ref ctfc) If no DW_AT_byte_size nor DW_AT_bit_size are defined, this function returns 0. */ -static uint32_t +static unsigned HOST_WIDE_INT ctf_die_bitsize (dw_die_ref die) { dw_attr_node *attr_byte_size = get_AT (die, DW_AT_byte_size); @@ -225,7 +221,9 @@ gen_ctf_base_type (ctf_container_ref ctfc, dw_die_ref type) ctf_encoding_t ctf_encoding = {0, 0, 0}; unsigned int encoding = get_AT_unsigned (type, DW_AT_encoding); - unsigned int bit_size = ctf_die_bitsize (type); + /* Bit size for the base types handled here should never be extremely large + (BITSIZE_MAXWIDTH at the upper end for _BitInt). */ + unsigned int bit_size = (unsigned int) ctf_die_bitsize (type); const char * name_string = get_AT_string (type, DW_AT_name); switch (encoding) @@ -514,7 +512,7 @@ gen_ctf_modifier_type (ctf_container_ref ctfc, dw_die_ref modifier) static ctf_dtdef_ref gen_ctf_sou_type (ctf_container_ref ctfc, dw_die_ref sou, uint32_t kind) { - uint32_t bit_size = ctf_die_bitsize (sou); + unsigned HOST_WIDE_INT bit_size = ctf_die_bitsize (sou); int declaration_p = get_AT_flag (sou, DW_AT_declaration); const char *sou_name = get_AT_string (sou, DW_AT_name); @@ -564,7 +562,7 @@ gen_ctf_sou_type (ctf_container_ref ctfc, dw_die_ref sou, uint32_t kind) { dw_attr_node *attr; HOST_WIDE_INT bitpos = 0; - HOST_WIDE_INT bitsize = ctf_die_bitsize (c); + unsigned HOST_WIDE_INT bitsize = ctf_die_bitsize (c); HOST_WIDE_INT bit_offset; /* The bit offset is given in bits and it may be @@ -581,7 +579,7 @@ gen_ctf_sou_type (ctf_container_ref ctfc, dw_die_ref sou, uint32_t kind) bitpos = field_location + bit_offset; else { - HOST_WIDE_INT bit_size; + unsigned HOST_WIDE_INT bit_size; attr = get_AT (c, DW_AT_byte_size); if (attr) @@ -730,7 +728,7 @@ static ctf_dtdef_ref 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 HOST_WIDE_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); diff --git a/gcc/testsuite/gcc.dg/debug/ctf/ctf-struct-3.c b/gcc/testsuite/gcc.dg/debug/ctf/ctf-struct-3.c new file mode 100644 index 000000000000..bf0be87c7c94 --- /dev/null +++ b/gcc/testsuite/gcc.dg/debug/ctf/ctf-struct-3.c @@ -0,0 +1,32 @@ +/* PR debug/121411. + Test for compilation of very large struct types. + The ctt_size for the struct shall encode CTF_LSIZE_SENT to indicate the + large struct encoding is used. */ + +/* { dg-do compile { target { lp64 || llp64 } } } */ +/* { dg-options "-O0 -gctf -dA" } */ + +struct huge +{ /* bit offset */ + unsigned char a1[0xffffffff]; /* 0 */ + unsigned char a2[0xffffffff]; /* 7fffffff8 */ + char x; /* ffffffff0 */ + char y; /* ffffffff8 */ + char z; /* 1000000000 */ +}; + +struct huge v; + +/* Verify struct is encoded with large type encoding format. */ +/* { dg-final { scan-assembler-times "0x1a000005\[\t \]+\[^\n\]*ctt_info" 1 } } */ +/* { dg-final { scan-assembler-times "0xffffffff\[\t \]+\[^\n\]*ctt_size" 1 } } */ +/* { dg-final { scan-assembler-times "0x2\[\t \]+\[^\n\]*ctt_lsizehi" 1 } } */ +/* { dg-final { scan-assembler-times "ctt_lsizelo" 1 } } */ + +/* Verify member offsets are correct for large offsets. */ +/* { dg-final { scan-assembler-times "0x7\[\t \]+\[^\n\]*ctlm_offsethi" 1 } } */ +/* { dg-final { scan-assembler-times "0xf\[\t \]+\[^\n\]*ctlm_offsethi" 2 } } */ +/* { dg-final { scan-assembler-times "0x10\[\t \]+\[^\n\]*ctlm_offsethi" 1 } } */ + +/* { dg-final { scan-assembler-times "0xfffffff8\[\t \]+\[^\n\]*ctlm_offsetlo" 2 } } */ +/* { dg-final { scan-assembler-times "0xfffffff0\[\t \]+\[^\n\]*ctlm_offsetlo" 1 } } */
