Skip to contentSkip to conversation

powerpc64/ftrace: fixes clang build issues

PATCH Applied Hari Bathini 3 participants 6 messages Jan 27 – Mar 11

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v2 cover powerpc64/ftrace: fixes clang build issues

Support for -fpatchable-function-entry on ppc64le was added in Clang with [1]. Faced a couple of issues while building the linux kernel with Clang though. Patches 1 & 2 address those issues. The last patch provides workaround to ensure, an open issue [2] in Clang with -fpatchable-function-entry support, does not impact the linux ftrace subsystem.

[1] https://github.com/llvm/llvm-project/pull/151569
[2] https://github.com/llvm/llvm-project/issues/163706

Hari Bathini (3):
powerpc64: make clang cross-build friendly powerpc64/ftrace: fix OOL stub count with clang powerpc64/ftrace: workaround clang recording GEP in
__patchable_function_entries

 arch/powerpc/Kconfig                          |  4 +--
 arch/powerpc/kernel/trace/ftrace.c            | 26 ++++++++++++++++---
 ....sh => check-fpatchable-function-entry.sh} |  0
 arch/powerpc/tools/ftrace-gen-ool-stubs.sh    |  4 +--
 4 files changed, 26 insertions(+), 8 deletions(-)
 rename arch/powerpc/tools/{gcc-check-fpatchable-function-entry.sh => check-fpatchable-function-entry.sh} (100%)
Signature · 1 line
2.52.0
#1 lore raw

PATCH v2 1/3 powerpc64: make clang cross-build friendly

ARCH_USING_PATCHABLE_FUNCTION_ENTRY depends on toolchain support for -fpatchable-function-entry option. The current script that checks for this support only handles GCC. Rename the script and extend it to detect support for -fpatchable-function-entry with Clang as well, allowing clean cross-compilation with Clang toolchains.

---

* Changes in v2:
- Use $(CLANG_FLAGS) instead of manually checking CC_IS_CLANG

 arch/powerpc/Kconfig                                          | 4 ++--
 ...e-function-entry.sh => check-fpatchable-function-entry.sh} | 0
 2 files changed, 2 insertions(+), 2 deletions(-)
 rename arch/powerpc/tools/{gcc-check-fpatchable-function-entry.sh => check-fpatchable-function-entry.sh} (100%)
arch/powerpc/Kconfig +2 -2
index b8d36a261009..dcb9894e2004 100644
@@ -571,8 +571,8 @@ config ARCH_USING_PATCHABLE_FUNCTION_ENTRY
 	depends on FUNCTION_TRACER && (PPC32 || PPC64_ELF_ABI_V2)
 	depends on $(cc-option,-fpatchable-function-entry=2)
 	def_bool y if PPC32
-	def_bool $(success,$(srctree)/arch/powerpc/tools/gcc-check-fpatchable-function-entry.sh $(CC) -mlittle-endian) if PPC64 && CPU_LITTLE_ENDIAN
-	def_bool $(success,$(srctree)/arch/powerpc/tools/gcc-check-fpatchable-function-entry.sh $(CC) -mbig-endian) if PPC64 && CPU_BIG_ENDIAN
+	def_bool $(success,$(srctree)/arch/powerpc/tools/check-fpatchable-function-entry.sh $(CC) $(CLANG_FLAGS) -mlittle-endian) if PPC64 && CPU_LITTLE_ENDIAN
+	def_bool $(success,$(srctree)/arch/powerpc/tools/check-fpatchable-function-entry.sh $(CC) -mbig-endian) if PPC64 && CPU_BIG_ENDIAN
 
 config PPC_FTRACE_OUT_OF_LINE
 	def_bool PPC64 && ARCH_USING_PATCHABLE_FUNCTION_ENTRY
arch/powerpc/tools/check-fpatchable-function-entry.sh renamed from arch/powerpc/tools/gcc-check-fpatchable-function-entry.sh +0 -0
similarity index 100%
rename from arch/powerpc/tools/gcc-check-fpatchable-function-entry.sh
rename to arch/powerpc/tools/check-fpatchable-function-entry.sh
Signature · 1 line
2.52.0
#2 lore raw

PATCH v2 2/3 powerpc64/ftrace: fix OOL stub count with clang

In reply to Hari Bathini

The total number of out-of-line (OOL) stubs required for function tracing is determined using the following command:

    $(OBJDUMP) -r -j __patchable_function_entries vmlinux.o

While this works correctly with GNU objdump, llvm-objdump does not list the expected relocation records for this section. Fix this by using the -d option and counting R_PPC64_ADDR64 relocation entries. This works as desired with both objdump and llvm-objdump.

---

* No changes since v1.

 arch/powerpc/tools/ftrace-gen-ool-stubs.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/powerpc/tools/ftrace-gen-ool-stubs.sh +2 -2
index bac186bdf64a..9218d43aeb54 100755
@@ -15,9 +15,9 @@ if [ -z "$is_64bit" ]; then
 	RELOCATION=R_PPC_ADDR32
 fi
 
-num_ool_stubs_total=$($objdump -r -j __patchable_function_entries "$vmlinux_o" |
+num_ool_stubs_total=$($objdump -r -j __patchable_function_entries -d "$vmlinux_o" |
 		      grep -c "$RELOCATION")
-num_ool_stubs_inittext=$($objdump -r -j __patchable_function_entries "$vmlinux_o" |
+num_ool_stubs_inittext=$($objdump -r -j __patchable_function_entries -d "$vmlinux_o" |
 			 grep -e ".init.text" -e ".text.startup" | grep -c "$RELOCATION")
 num_ool_stubs_text=$((num_ool_stubs_total - num_ool_stubs_inittext))
 
Signature · 1 line
2.52.0
#3 lore raw

PATCH v2 3/3 powerpc64/ftrace: workaround clang recording GEP in __patchable_function_entries

In reply to Hari Bathini

Support for -fpatchable-function-entry on ppc64le was added in Clang with [1]. However, when no prefix NOPs are specified - as is the case with CONFIG_PPC_FTRACE_OUT_OF_LINE - the first NOP is emitted at LEP, but Clang records the Global Entry Point (GEP) unlike GCC which does record the Local Entry Point (LEP). Issue [2] has been raised to align Clang's behavior with GCC. As a temporary workaround to ensure ftrace initialization works as expected with Clang, derive the LEP using ppc_function_entry() for kernel symbols and by looking for the below module GEP sequence for module addresses, until [2] is resolved:

	ld	r2, -8(r12)
	add	r2, r2, r12

[1] https://github.com/llvm/llvm-project/pull/151569
[2] https://github.com/llvm/llvm-project/issues/163706

Cc Maryam Moghadas <[email protected]>

---

* No changes since v1.

 arch/powerpc/kernel/trace/ftrace.c | 26 ++++++++++++++++++++++----
 1 file changed, 22 insertions(+), 4 deletions(-)
arch/powerpc/kernel/trace/ftrace.c +22 -4
index 841d077e2825..1b2f293e7dcb 100644
@@ -37,11 +37,29 @@ unsigned long ftrace_call_adjust(unsigned long addr)
 	if (addr >= (unsigned long)__exittext_begin && addr < (unsigned long)__exittext_end)
 		return 0;
 
-	if (IS_ENABLED(CONFIG_ARCH_USING_PATCHABLE_FUNCTION_ENTRY) &&
-	    !IS_ENABLED(CONFIG_PPC_FTRACE_OUT_OF_LINE)) {
-		addr += MCOUNT_INSN_SIZE;
-		if (IS_ENABLED(CONFIG_DYNAMIC_FTRACE_WITH_CALL_OPS))
+	if (IS_ENABLED(CONFIG_ARCH_USING_PATCHABLE_FUNCTION_ENTRY)) {
+		if (!IS_ENABLED(CONFIG_PPC_FTRACE_OUT_OF_LINE)) {
 			addr += MCOUNT_INSN_SIZE;
+			if (IS_ENABLED(CONFIG_DYNAMIC_FTRACE_WITH_CALL_OPS))
+				addr += MCOUNT_INSN_SIZE;
+		} else if (IS_ENABLED(CONFIG_CC_IS_CLANG) && IS_ENABLED(CONFIG_PPC64)) {
+			/*
+			 * addr points to global entry point though the NOP was emitted at local
+			 * entry point due to https://github.com/llvm/llvm-project/issues/163706
+			 * Handle that here with ppc_function_entry() for kernel symbols while
+			 * adjusting module addresses in the else case, by looking for the below
+			 * module global entry point sequence:
+			 *	ld    r2, -8(r12)
+			 *	add   r2, r2, r12
+			 */
+			if (is_kernel_text(addr) || is_kernel_inittext(addr))
+				addr = ppc_function_entry((void *)addr);
+			else if ((ppc_inst_val(ppc_inst_read((u32 *)addr)) ==
+				  PPC_RAW_LD(_R2, _R12, -8)) &&
+				 (ppc_inst_val(ppc_inst_read((u32 *)(addr+4))) ==
+				  PPC_RAW_ADD(_R2, _R2, _R12)))
+				addr += 8;
+		}
 	}
 
 	return addr;
Signature · 1 line
2.52.0
Tested-by on the v2 series
#4 lore raw
In reply to Hari Bathini
Quoted 17 lines · Hari Bathini

Support for -fpatchable-function-entry on ppc64le was added in Clang with [1]. Faced a couple of issues while building the linux kernel with Clang though. Patches 1 & 2 address those issues. The last patch provides workaround to ensure, an open issue [2] in Clang with -fpatchable-function-entry support, does not impact the linux ftrace subsystem.

[1] https://github.com/llvm/llvm-project/pull/151569
[2] https://github.com/llvm/llvm-project/issues/163706

Hari Bathini (3):
powerpc64: make clang cross-build friendly powerpc64/ftrace: fix OOL stub count with clang powerpc64/ftrace: workaround clang recording GEP in
__patchable_function_entries

Quoted 7 lines
  arch/powerpc/Kconfig                          |  4 +--
  arch/powerpc/kernel/trace/ftrace.c            | 26 ++++++++++++++++---
  ....sh => check-fpatchable-function-entry.sh} |  0
  arch/powerpc/tools/ftrace-gen-ool-stubs.sh    |  4 +--
  4 files changed, 26 insertions(+), 8 deletions(-)

rename arch/powerpc/tools/{gcc-check-fpatchable-function-entry.sh => check-fpatchable-function-entry.sh} (100%)

Tested this patch series, and bulild is successful with both GCC and clang23.x version.

KERNEL PARAMETER enabled:

CONFIG_ARCH_USING_PATCHABLE_FUNCTION_ENTRY=y
CONFIG_PPC_FTRACE_OUT_OF_LINE=y
CONFIG_PPC_FTRACE_OUT_OF_LINE_NUM_RESERVE=32768

clang version 23.0.0git

Please add below tag for the series.

Regards,

Venkat.

maintainer Applied
#5 lore raw
In reply to Hari Bathini
Quoted 8 lines · Hari Bathini

Support for -fpatchable-function-entry on ppc64le was added in Clang with [1]. Faced a couple of issues while building the linux kernel with Clang though. Patches 1 & 2 address those issues. The last patch provides workaround to ensure, an open issue [2] in Clang with -fpatchable-function-entry support, does not impact the linux ftrace subsystem.

[...]

Applied to powerpc/fixes.

[1/3] powerpc64: make clang cross-build friendly

      https://git.kernel.org/powerpc/c/73cdf24e81e4eba52a40a6b10c6cf285d0ac23fd
[2/3] powerpc64/ftrace: fix OOL stub count with clang
      https://git.kernel.org/powerpc/c/875612a7745013a43c67493cb0583ee3f7476344
[3/3] powerpc64/ftrace: workaround clang recording GEP in __patchable_function_entries
      https://git.kernel.org/powerpc/c/db54c28702f7270e74dce36c84cb0db4cec96389

cheers