Skip to contentSkip to conversation

lib/Kconfig.debug: Add check for non-constant .{s,u}leb128 support to DWARF5

PATCH Superseded Nathan Chancellor 3 participants 7 messages Sep 28, 2022 – Oct 3, 2022

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Nathan Chancellor

author maintainer
#0 lore raw

PATCH 1/1 lib/Kconfig.debug: Add check for non-constant .{s,u}leb128 support to DWARF5

When building with a RISC-V kernel with DWARF5 debug info using clang and the GNU assembler, several instances of the following error appear:

/tmp/vgettimeofday-48aa35.s:2963: Error: non-constant .uleb128 is not supported

Dumping the .s file reveals these .uleb128 directives come from .debug_loc and .debug_ranges:

.Ldebug_loc0:

          .byte   4                               # DW_LLE_offset_pair
          .uleb128 .Lfunc_begin0-.Lfunc_begin0    #   starting offset
          .uleb128 .Ltmp1-.Lfunc_begin0           #   ending offset
          .byte   1                               # Loc expr size
          .byte   90                              # DW_OP_reg10
          .byte   0                               # DW_LLE_end_of_list

.Ldebug_ranges0:

          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp6-.Lfunc_begin0           #   starting offset
          .uleb128 .Ltmp27-.Lfunc_begin0          #   ending offset
          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp28-.Lfunc_begin0          #   starting offset
          .uleb128 .Ltmp30-.Lfunc_begin0          #   ending offset
          .byte   0                               # DW_RLE_end_of_list

There is an outstanding binutils issue to support a non-constant operand to .sleb128 and .uleb128 in GAS for RISC-V but there does not appear to be any movement on it, due to concerns over how it would work with linker relaxation.

To avoid these build errors, prevent DWARF5 from being selected when using clang and an assembler that does not have support for these symbol deltas, which can be easily checked in Kconfig with as-instr plus the small test program from the dwz test suite from the binutils issue.

---
 lib/Kconfig.debug | 7 +++++++
 1 file changed, 7 insertions(+)
lib/Kconfig.debug +7 -0
index d3e5f36bb01e..19de03ead2ed 100644
@@ -231,6 +231,9 @@ config DEBUG_INFO
 	  in the "Debug information" choice below, indicating that debug
 	  information will be generated for build targets.
 
+config AS_HAS_NON_CONST_LEB128
+	def_bool $(as-instr,.uleb128 .Lexpr_end4 - .Lexpr_start3\n.Lexpr_start3:\n.Lexpr_end4:)
+
 choice
 	prompt "Debug information"
 	depends on DEBUG_KERNEL
@@ -277,6 +280,10 @@ config DEBUG_INFO_DWARF5
 	bool "Generate DWARF Version 5 debuginfo"
 	select DEBUG_INFO
 	depends on !CC_IS_CLANG || (CC_IS_CLANG && (AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502)))
+	# Clang is known to generate .{s,u}leb128 with symbol deltas with
+	# DWARF5, which some targets may not support.
+	# https://sourceware.org/bugzilla/show_bug.cgi?id=27215
+	depends on !CC_IS_CLANG || AS_HAS_NON_CONST_LEB128
 	help
 	  Generate DWARF v5 debug info. Requires binutils 2.35.2, gcc 5.0+ (gcc
 	  5.0+ accepts the -gdwarf-5 flag but only had partial support for some

base-commit: f76349cf41451c5c42a99f18a9163377e4b364ff

Signature · 1 line
2.37.3
#1 lore raw
Quoted 59 lines · Nathan Chancellor

When building with a RISC-V kernel with DWARF5 debug info using clang and the GNU assembler, several instances of the following error appear:

/tmp/vgettimeofday-48aa35.s:2963: Error: non-constant .uleb128 is not supported

Dumping the .s file reveals these .uleb128 directives come from .debug_loc and .debug_ranges:

.Ldebug_loc0:

          .byte   4                               # DW_LLE_offset_pair
          .uleb128 .Lfunc_begin0-.Lfunc_begin0    #   starting offset
          .uleb128 .Ltmp1-.Lfunc_begin0           #   ending offset
          .byte   1                               # Loc expr size
          .byte   90                              # DW_OP_reg10
          .byte   0                               # DW_LLE_end_of_list

.Ldebug_ranges0:

          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp6-.Lfunc_begin0           #   starting offset
          .uleb128 .Ltmp27-.Lfunc_begin0          #   ending offset
          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp28-.Lfunc_begin0          #   starting offset
          .uleb128 .Ltmp30-.Lfunc_begin0          #   ending offset
          .byte   0                               # DW_RLE_end_of_list

There is an outstanding binutils issue to support a non-constant operand to .sleb128 and .uleb128 in GAS for RISC-V but there does not appear to be any movement on it, due to concerns over how it would work with linker relaxation.

To avoid these build errors, prevent DWARF5 from being selected when using clang and an assembler that does not have support for these symbol deltas, which can be easily checked in Kconfig with as-instr plus the small test program from the dwz test suite from the binutils issue.

---
 lib/Kconfig.debug | 7 +++++++
 1 file changed, 7 insertions(+)
lib/Kconfig.debug +3 -0
index d3e5f36bb01e..19de03ead2ed 100644
@@ -231,6 +231,9 @@ config DEBUG_INFO
          in the "Debug information" choice below, indicating that debug
          information will be generated for build targets.
 
+config AS_HAS_NON_CONST_LEB128
+       def_bool $(as-instr,.uleb128 .Lexpr_end4 - .Lexpr_start3\n.Lexpr_start3:\n.Lexpr_end4:)
+
 choice
        prompt "Debug information"
        depends on DEBUG_KERNEL

lib/Kconfig.debug config DEBUG_INFO_DWARF5 · line 286

Quoted 7 lines
lib/Kconfig.debug +4 -0
@@ -277,6 +280,10 @@ config DEBUG_INFO_DWARF5
        bool "Generate DWARF Version 5 debuginfo"
        select DEBUG_INFO
        depends on !CC_IS_CLANG || (CC_IS_CLANG && (AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502)))
+       # Clang is known to generate .{s,u}leb128 with symbol deltas with
+       # DWARF5, which some targets may not support.
+       # https://sourceware.org/bugzilla/show_bug.cgi?id=27215
+       depends on !CC_IS_CLANG || AS_HAS_NON_CONST_LEB128

Reraising my concern from
https://github.com/ClangBuiltLinux/linux/issues/1719#issuecomment-1258678969

We've put a fair amount of work into getting CC=clang LLVM_IAS=0 to work for DWARF v5 (both on the GNU binutils side, and Kbuild), I'd hate to see that effectively knee-capped because of an issue in GNU binutils that is only relevant for one architecture.

I'd concede support for ARCH=riscv, but not for all other architectures, which this effectively does.

Quoted 9 lines
lib/Kconfig.debug +0 -0
        help
          Generate DWARF v5 debug info. Requires binutils 2.35.2, gcc 5.0+ (gcc
          5.0+ accepts the -gdwarf-5 flag but only had partial support for some

base-commit: f76349cf41451c5c42a99f18a9163377e4b364ff --
2.37.3

Signature · 2 lines
Thanks,
~Nick Desaulniers
author maintainer
#2 lore raw
Quoted 62 lines · Nick Desaulniers
Quoted 61 lines · Nathan Chancellor

When building with a RISC-V kernel with DWARF5 debug info using clang and the GNU assembler, several instances of the following error appear:

/tmp/vgettimeofday-48aa35.s:2963: Error: non-constant .uleb128 is not supported

Dumping the .s file reveals these .uleb128 directives come from .debug_loc and .debug_ranges:

.Ldebug_loc0:

          .byte   4                               # DW_LLE_offset_pair
          .uleb128 .Lfunc_begin0-.Lfunc_begin0    #   starting offset
          .uleb128 .Ltmp1-.Lfunc_begin0           #   ending offset
          .byte   1                               # Loc expr size
          .byte   90                              # DW_OP_reg10
          .byte   0                               # DW_LLE_end_of_list

.Ldebug_ranges0:

          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp6-.Lfunc_begin0           #   starting offset
          .uleb128 .Ltmp27-.Lfunc_begin0          #   ending offset
          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp28-.Lfunc_begin0          #   starting offset
          .uleb128 .Ltmp30-.Lfunc_begin0          #   ending offset
          .byte   0                               # DW_RLE_end_of_list

There is an outstanding binutils issue to support a non-constant operand to .sleb128 and .uleb128 in GAS for RISC-V but there does not appear to be any movement on it, due to concerns over how it would work with linker relaxation.

To avoid these build errors, prevent DWARF5 from being selected when using clang and an assembler that does not have support for these symbol deltas, which can be easily checked in Kconfig with as-instr plus the small test program from the dwz test suite from the binutils issue.

---
 lib/Kconfig.debug | 7 +++++++
 1 file changed, 7 insertions(+)
lib/Kconfig.debug +3 -0
index d3e5f36bb01e..19de03ead2ed 100644
@@ -231,6 +231,9 @@ config DEBUG_INFO
          in the "Debug information" choice below, indicating that debug
          information will be generated for build targets.
 
+config AS_HAS_NON_CONST_LEB128
+       def_bool $(as-instr,.uleb128 .Lexpr_end4 - .Lexpr_start3\n.Lexpr_start3:\n.Lexpr_end4:)
+
 choice
        prompt "Debug information"
        depends on DEBUG_KERNEL
@@ -277,6 +280,10 @@ config DEBUG_INFO_DWARF5
        bool "Generate DWARF Version 5 debuginfo"
        select DEBUG_INFO
Quoted 8 lines

lib/Kconfig.debug config DEBUG_INFO_DWARF5 · line 286

Quoted 5 lines
lib/Kconfig.debug +4 -0
        depends on !CC_IS_CLANG || (CC_IS_CLANG && (AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502)))
+       # Clang is known to generate .{s,u}leb128 with symbol deltas with
+       # DWARF5, which some targets may not support.
+       # https://sourceware.org/bugzilla/show_bug.cgi?id=27215
+       depends on !CC_IS_CLANG || AS_HAS_NON_CONST_LEB128

Reraising my concern from
https://github.com/ClangBuiltLinux/linux/issues/1719#issuecomment-1258678969

Sorry, I thought I addressed your concern with my comment right below it but I probably should have worded it better.

Quoted 4 lines

We've put a fair amount of work into getting CC=clang LLVM_IAS=0 to work for DWARF v5 (both on the GNU binutils side, and Kbuild), I'd hate to see that effectively knee-capped because of an issue in GNU binutils that is only relevant for one architecture.

Sure, that is a completely reasonable concern. However...

Quoted 2 lines

I'd concede support for ARCH=riscv, but not for all other architectures, which this effectively does.

No, it does not, CONFIG_AS_HAS_NON_CONST_LEB128 can still be enabled when GNU as supports this construct for a particular architecture; as far as I can tell, RISC-V is the only one that doesn't. See the tests with ARCH=arm64 and ARCH=x86_64 compared with ARCH=riscv below.

$ make -skj"$(nproc)" ARCH=x86_64 LLVM=1 LLVM_IAS=0 defconfig
$ rg "CONFIG_AS_(IS|VERSION|HAS)" .config
9:CONFIG_AS_IS_GNU=y
10:CONFIG_AS_VERSION=23950
4750:CONFIG_AS_HAS_NON_CONST_LEB128=y
$ make -skj"$(nproc)" ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- LLVM=1 LLVM_IAS=0 defconfig
$ rg "CONFIG_AS_(IS|VERSION|HAS)" .config
9:CONFIG_AS_IS_GNU=y
10:CONFIG_AS_VERSION=23950
442:CONFIG_AS_HAS_LDAPR=y
443:CONFIG_AS_HAS_LSE_ATOMICS=y
451:CONFIG_AS_HAS_ARMV8_2=y
452:CONFIG_AS_HAS_SHA3=y
465:CONFIG_AS_HAS_PAC=y
466:CONFIG_AS_HAS_CFI_NEGATE_RA_STATE=y
473:CONFIG_AS_HAS_ARMV8_4=y
480:CONFIG_AS_HAS_ARMV8_5=y
9719:CONFIG_AS_HAS_NON_CONST_LEB128=y
$ make -skj"$(nproc)" ARCH=riscv CROSS_COMPILE=riscv64-linux-gnu- LLVM=1 LLVM_IAS=0 defconfig
$ rg "CONFIG_AS_(IS|VERSION|HAS)" .config
9:CONFIG_AS_IS_GNU=y
10:CONFIG_AS_VERSION=23950

Cheers,
Nathan

Reviewed-by
#3 lore raw
Quoted 7 lines · Nathan Chancellor
Quoted 2 lines · Nick Desaulniers

Reraising my concern from
https://github.com/ClangBuiltLinux/linux/issues/1719#issuecomment-1258678969

Sorry, I thought I addressed your concern with my comment right below it but I probably should have worded it better.

No, I just missed your point about other architectures.

Quoted 7 lines
Quoted 4 lines

We've put a fair amount of work into getting CC=clang LLVM_IAS=0 to work for DWARF v5 (both on the GNU binutils side, and Kbuild), I'd hate to see that effectively knee-capped because of an issue in GNU binutils that is only relevant for one architecture.

Sure, that is a completely reasonable concern. However...

Quoted 7 lines
Quoted 2 lines

I'd concede support for ARCH=riscv, but not for all other architectures, which this effectively does.

No, it does not, CONFIG_AS_HAS_NON_CONST_LEB128 can still be enabled when GNU as supports this construct for a particular architecture; as far as I can tell, RISC-V is the only one that doesn't. See the tests with ARCH=arm64 and ARCH=x86_64 compared with ARCH=riscv below.

Ah, sorry for the noise then. Thanks for the patch.

Signature · 2 lines
Thanks,
~Nick Desaulniers
#4 lore raw
Quoted 27 lines · Nick Desaulniers
Quoted 7 lines · Nathan Chancellor
Quoted 2 lines · Nick Desaulniers

Reraising my concern from
https://github.com/ClangBuiltLinux/linux/issues/1719#issuecomment-1258678969

Sorry, I thought I addressed your concern with my comment right below it but I probably should have worded it better.

No, I just missed your point about other architectures.

Quoted 7 lines
Quoted 4 lines

We've put a fair amount of work into getting CC=clang LLVM_IAS=0 to work for DWARF v5 (both on the GNU binutils side, and Kbuild), I'd hate to see that effectively knee-capped because of an issue in GNU binutils that is only relevant for one architecture.

Sure, that is a completely reasonable concern. However...

Quoted 7 lines
Quoted 2 lines

I'd concede support for ARCH=riscv, but not for all other architectures, which this effectively does.

No, it does not, CONFIG_AS_HAS_NON_CONST_LEB128 can still be enabled when GNU as supports this construct for a particular architecture; as far as I can tell, RISC-V is the only one that doesn't. See the tests with ARCH=arm64 and ARCH=x86_64 compared with ARCH=riscv below.

Quoted 6 lines

Ah, sorry for the noise then. Thanks for the patch.

--
Thanks,
~Nick Desaulniers

This patch is incomplete.

It looks like Clang 14 switched to DWARF v5 by default.

I see the same errors for
CONFIG_DEBUG_INFO_DWARF_TOOLCHAIN_DEFAULT=y

masahiro@zoe:~/ref/linux$ clang --version | head -n1
Ubuntu clang version 14.0.0-1ubuntu1
masahiro@zoe:~/ref/linux$ grep DEBUG_INFO_DWARF .config
CONFIG_DEBUG_INFO_DWARF_TOOLCHAIN_DEFAULT=y
# CONFIG_DEBUG_INFO_DWARF4 is not set
# CONFIG_DEBUG_INFO_DWARF5 is not set
masahiro@zoe:~/ref/linux$ make ARCH=riscv LLVM=1 LLVM_IAS=0
CROSS_COMPILE=riscv64-linux-gnu- -j24
  CALL    scripts/atomic/check-atomics.sh
  CALL    scripts/checksyscalls.sh
  CC      arch/riscv/kernel/vdso/vgettimeofday.o
/tmp/vgettimeofday-5997b4.s: Assembler messages:
/tmp/vgettimeofday-5997b4.s:2698: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2699: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2705: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2706: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2712: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2713: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2719: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2720: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2726: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2727: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2731: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2732: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2736: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2737: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2743: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2744: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2748: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2749: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2755: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2756: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2760: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2761: Error: non-constant .uleb128 is not supported
/tmp/vgettimeofday-5997b4.s:2765: Error: non-constant .uleb128 is not supported
   ...
Signature · 2 lines
Best Regards
Masahiro Yamada
#5 lore raw
In reply to Nathan Chancellor
Quoted 59 lines · Nathan Chancellor

When building with a RISC-V kernel with DWARF5 debug info using clang and the GNU assembler, several instances of the following error appear:

/tmp/vgettimeofday-48aa35.s:2963: Error: non-constant .uleb128 is not supported

Dumping the .s file reveals these .uleb128 directives come from .debug_loc and .debug_ranges:

.Ldebug_loc0:

          .byte   4                               # DW_LLE_offset_pair
          .uleb128 .Lfunc_begin0-.Lfunc_begin0    #   starting offset
          .uleb128 .Ltmp1-.Lfunc_begin0           #   ending offset
          .byte   1                               # Loc expr size
          .byte   90                              # DW_OP_reg10
          .byte   0                               # DW_LLE_end_of_list

.Ldebug_ranges0:

          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp6-.Lfunc_begin0           #   starting offset
          .uleb128 .Ltmp27-.Lfunc_begin0          #   ending offset
          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp28-.Lfunc_begin0          #   starting offset
          .uleb128 .Ltmp30-.Lfunc_begin0          #   ending offset
          .byte   0                               # DW_RLE_end_of_list

There is an outstanding binutils issue to support a non-constant operand to .sleb128 and .uleb128 in GAS for RISC-V but there does not appear to be any movement on it, due to concerns over how it would work with linker relaxation.

To avoid these build errors, prevent DWARF5 from being selected when using clang and an assembler that does not have support for these symbol deltas, which can be easily checked in Kconfig with as-instr plus the small test program from the dwz test suite from the binutils issue.

---
 lib/Kconfig.debug | 7 +++++++
 1 file changed, 7 insertions(+)
lib/Kconfig.debug +3 -0
index d3e5f36bb01e..19de03ead2ed 100644
@@ -231,6 +231,9 @@ config DEBUG_INFO
          in the "Debug information" choice below, indicating that debug
          information will be generated for build targets.
 
+config AS_HAS_NON_CONST_LEB128
+       def_bool $(as-instr,.uleb128 .Lexpr_end4 - .Lexpr_start3\n.Lexpr_start3:\n.Lexpr_end4:)
+
 choice
        prompt "Debug information"
        depends on DEBUG_KERNEL

lib/Kconfig.debug config DEBUG_INFO_DWARF5 · line 285

Quoted 6 lines
lib/Kconfig.debug +3 -0
@@ -277,6 +280,10 @@ config DEBUG_INFO_DWARF5
        bool "Generate DWARF Version 5 debuginfo"
        select DEBUG_INFO
        depends on !CC_IS_CLANG || (CC_IS_CLANG && (AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502)))
+       # Clang is known to generate .{s,u}leb128 with symbol deltas with
+       # DWARF5, which some targets may not support.
+       # https://sourceware.org/bugzilla/show_bug.cgi?id=27215

If you plan to patch both DWARF_TOOLCHAIN_DEFAULT and DWARF5, it will be cleaner to move this comment to AS_HAS_NON_CONST_LEB128.

lib/Kconfig.debug config DEBUG_INFO_DWARF5 · line 286

Quoted 1 line
lib/Kconfig.debug +1 -0
+       depends on !CC_IS_CLANG || AS_HAS_NON_CONST_LEB128

The condition "!CC_IS_CLANG" is repeated here.

If you use the following patch as basic,
https://lore.kernel.org/lkml/[email protected]/T/#u

you can write the code like this:

!CC_IS_CLANG || AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502 && AS_HAS_NON_CONST_LEB128)

Another big hammer solution is to give up Clang+GAS for CONFIG_DEBUG_INFO. If we go this way, this patch is unneeded, though.
Thoughts?

Signature · 2 lines
Best Regards
Masahiro Yamada
author maintainer
#6 lore raw
Quoted 53 lines · Masahiro Yamada
Quoted 52 lines · Nathan Chancellor

When building with a RISC-V kernel with DWARF5 debug info using clang and the GNU assembler, several instances of the following error appear:

/tmp/vgettimeofday-48aa35.s:2963: Error: non-constant .uleb128 is not supported

Dumping the .s file reveals these .uleb128 directives come from .debug_loc and .debug_ranges:

.Ldebug_loc0:

          .byte   4                               # DW_LLE_offset_pair
          .uleb128 .Lfunc_begin0-.Lfunc_begin0    #   starting offset
          .uleb128 .Ltmp1-.Lfunc_begin0           #   ending offset
          .byte   1                               # Loc expr size
          .byte   90                              # DW_OP_reg10
          .byte   0                               # DW_LLE_end_of_list

.Ldebug_ranges0:

          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp6-.Lfunc_begin0           #   starting offset
          .uleb128 .Ltmp27-.Lfunc_begin0          #   ending offset
          .byte   4                               # DW_RLE_offset_pair
          .uleb128 .Ltmp28-.Lfunc_begin0          #   starting offset
          .uleb128 .Ltmp30-.Lfunc_begin0          #   ending offset
          .byte   0                               # DW_RLE_end_of_list

There is an outstanding binutils issue to support a non-constant operand to .sleb128 and .uleb128 in GAS for RISC-V but there does not appear to be any movement on it, due to concerns over how it would work with linker relaxation.

To avoid these build errors, prevent DWARF5 from being selected when using clang and an assembler that does not have support for these symbol deltas, which can be easily checked in Kconfig with as-instr plus the small test program from the dwz test suite from the binutils issue.

---
 lib/Kconfig.debug | 7 +++++++
 1 file changed, 7 insertions(+)
lib/Kconfig.debug +0 -0
index d3e5f36bb01e..19de03ead2ed 100644
@@ -231,6 +231,9 @@ config DEBUG_INFO
          in the "Debug information" choice below, indicating that debug
          information will be generated for build targets.
 
Quoted 16 lines

lib/Kconfig.debug config DEBUG_INFO_DWARF5 · line 285

Quoted 13 lines
lib/Kconfig.debug +6 -0
+config AS_HAS_NON_CONST_LEB128
+       def_bool $(as-instr,.uleb128 .Lexpr_end4 - .Lexpr_start3\n.Lexpr_start3:\n.Lexpr_end4:)
+
 choice
        prompt "Debug information"
        depends on DEBUG_KERNEL
@@ -277,6 +280,10 @@ config DEBUG_INFO_DWARF5
        bool "Generate DWARF Version 5 debuginfo"
        select DEBUG_INFO
        depends on !CC_IS_CLANG || (CC_IS_CLANG && (AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502)))
+       # Clang is known to generate .{s,u}leb128 with symbol deltas with
+       # DWARF5, which some targets may not support.
+       # https://sourceware.org/bugzilla/show_bug.cgi?id=27215

If you plan to patch both DWARF_TOOLCHAIN_DEFAULT and DWARF5, it will be cleaner to move this comment to AS_HAS_NON_CONST_LEB128.

Sure, that sounds reasonable! I can base this change on the series that you recently sent:

https://lore.kernel.org/[email protected]/

Quoted 11 lines

lib/Kconfig.debug config DEBUG_INFO_DWARF5 · line 286

Quoted 1 line
lib/Kconfig.debug +1 -0
+       depends on !CC_IS_CLANG || AS_HAS_NON_CONST_LEB128

The condition "!CC_IS_CLANG" is repeated here.

If you use the following patch as basic,
https://lore.kernel.org/lkml/[email protected]/T/#u

you can write the code like this:

!CC_IS_CLANG || AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502 && AS_HAS_NON_CONST_LEB128)

Right, I had initially had something along this line but the length of the line bothered some folks in pre-review so I opted for a second line. With your clean ups, it seems reasonable to move it back to the original line.

Quoted 3 lines

Another big hammer solution is to give up Clang+GAS for CONFIG_DEBUG_INFO. If we go this way, this patch is unneeded, though.
Thoughts?

I think this is a simple enough solution to avoid that big hammer at the moment but if we continue to run into corner cases like this, that is certainly worth considering.

Cheers,
Nathan