Skip to contentSkip to conversation

Rust support for powerpc

PATCH Superseded Mukesh Kumar Chaurasiya (IBM) 6 participants 19 messages Feb 10 – Mar 26

Subsystems (4)

v6 of 17 Superseded Latest is v17 →

lore
Share & tools
lore
#0 lore raw

v6 cover Rust support for powerpc

Enable experimental rust support for ppc64le and ppc32be. The patch for ppc32 has been provided by Link Mauve[1] and ppc64le support[2] has been merged over it. ppc32 needs some toolchain fixes mentioned in the patch rust: Add PowerPC support and the discussion for that is done here[1].

This has been tested on powernv9 hardware and power10 pseries qemu. I I request Link to test the ppc32 part as i don't have a hardware to test it out.

[1] https://lore.kernel.org/all/[email protected]
[2] https://lore.kernel.org/all/[email protected]

Changelog:
V5 -> V6:
- Added a missing Tested by from Venkat which got missed since V3
- Support is marked as Maintained instead of experimental
V5: https://lore.kernel.org/all/[email protected]

V4 -> V5:
- Removed a nested ifdef from PPC64 for Little endian toolchain
V4: https://lore.kernel.org/all/[email protected]

V3 -> V4:
- Co-developed-by header added in patch 1
V3: https://lore.kernel.org/all/[email protected]

V2 -> V3:
- Splited HAVE_RUST in 2 lines
- BINDGEN_TARGET_powerpc initialized before assigning the same to BINDGEN_TARGET
V2: https://lore.kernel.org/all/[email protected]

V1 -> V2:
- jump label fix for rust has been moved to a separate patch
- PPC32 support has been taken
- rust support has been marked experimental
- target.json dependency has been removed
- HAVE_RUST now depends on CPU_LITTLE_ENDIAN for PPC64

Link Mauve (1):
rust: Add PowerPC support

Mukesh Kumar Chaurasiya (IBM) (2):
powerpc/jump_label: adjust inline asm to be consistent
powerpc: Enable Rust for ppc64le

 Documentation/rust/arch-support.rst   |  1 +
 arch/powerpc/Kconfig                  |  2 ++
 arch/powerpc/Makefile                 |  7 +++++++
 arch/powerpc/include/asm/jump_label.h | 23 +++++++++++++----------
 rust/Makefile                         | 10 +++++++++-
 5 files changed, 32 insertions(+), 11 deletions(-)
Signature · 1 line
2.53.0
#1 lore raw

PATCH v6 1/3 powerpc/jump_label: adjust inline asm to be consistent

Added support for a new macro ARCH_STATIC_BRANCH_ASM in powerpc
to avoid duplication of inline asm between C and Rust. This is
inline with commit aecaf181651c '("jump_label: adjust inline asm to be consistent")'

---
 arch/powerpc/include/asm/jump_label.h | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)
arch/powerpc/include/asm/jump_label.h +13 -10
index d4eaba459a0e..a6b211502bfe 100644
@@ -15,14 +15,20 @@
 #define JUMP_ENTRY_TYPE		stringify_in_c(FTR_ENTRY_LONG)
 #define JUMP_LABEL_NOP_SIZE	4
 
+#define JUMP_TABLE_ENTRY(key, label)			\
+	".pushsection __jump_table,  \"aw\"	\n\t"	\
+	".long 1b - ., " label " - .		\n\t"	\
+	JUMP_ENTRY_TYPE key " - .		\n\t"	\
+	".popsection 				\n\t"
+
+#define ARCH_STATIC_BRANCH_ASM(key, label)		\
+	"1:	nop				\n\t"	\
+	JUMP_TABLE_ENTRY(key,label)
+
 static __always_inline bool arch_static_branch(struct static_key *key, bool branch)
 {
-	asm goto("1:\n\t"
-		 "nop # arch_static_branch\n\t"
-		 ".pushsection __jump_table,  \"aw\"\n\t"
-		 ".long 1b - ., %l[l_yes] - .\n\t"
-		 JUMP_ENTRY_TYPE "%c0 - .\n\t"
-		 ".popsection \n\t"
+	asm goto(
+		 ARCH_STATIC_BRANCH_ASM("%c0", "%l[l_yes]")
 		 : :  "i" (&((char *)key)[branch]) : : l_yes);
 
 	return false;
@@ -34,10 +40,7 @@ static __always_inline bool arch_static_branch_jump(struct static_key *key, bool
 {
 	asm goto("1:\n\t"
 		 "b %l[l_yes] # arch_static_branch_jump\n\t"
-		 ".pushsection __jump_table,  \"aw\"\n\t"
-		 ".long 1b - ., %l[l_yes] - .\n\t"
-		 JUMP_ENTRY_TYPE "%c0 - .\n\t"
-		 ".popsection \n\t"
+		 JUMP_TABLE_ENTRY("%c0", "%l[l_yes]")
 		 : :  "i" (&((char *)key)[branch]) : : l_yes);
 
 	return false;
Signature · 1 line
2.53.0
#2 lore raw

PATCH v6 2/3 rust: Add PowerPC support

In reply to Mukesh Kumar Chaurasiya (IBM)

From: Link Mauve <[email protected]>

For now only Big Endian 32-bit PowerPC is supported, as that is the only hardware I have. This has been tested on the Nintendo Wii so far, but I plan on also using it on the GameCube, Wii U and Apple G4.

These changes aren’t the only ones required to get the kernel to compile and link on PowerPC, libcore will also have to be changed to not use integer division to format u64, u128 and core::time::Duration, otherwise __udivdi3() and __umoddi3() will have to be added. I have tested this change by replacing the three implementations with unimplemented!() and it linked just fine.

---
 Documentation/rust/arch-support.rst | 1 +
 arch/powerpc/Kconfig                | 1 +
 arch/powerpc/Makefile               | 2 ++
 rust/Makefile                       | 4 +++-
 4 files changed, 7 insertions(+), 1 deletion(-)
Documentation/rust/arch-support.rst +1 -0
index 6e6a515d0899..70b9e192a7a0 100644
@@ -18,6 +18,7 @@ Architecture   Level of support  Constraints
 ``arm``        Maintained        ARMv7 Little Endian only.
 ``arm64``      Maintained        Little Endian only.
 ``loongarch``  Maintained        \-
+``powerpc``    Maintained        32-bit Big Endian only.
 ``riscv``      Maintained        ``riscv64`` and LLVM/Clang only.
 ``um``         Maintained        \-
 ``x86``        Maintained        ``x86_64`` only.
arch/powerpc/Kconfig +1 -0
index 9537a61ebae0..17db23b82e91 100644
@@ -283,6 +283,7 @@ config PPC
 	select HAVE_REGS_AND_STACK_ACCESS_API
 	select HAVE_RELIABLE_STACKTRACE
 	select HAVE_RSEQ
+	select HAVE_RUST			if PPC32
 	select HAVE_SAMPLE_FTRACE_DIRECT	if HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
 	select HAVE_SAMPLE_FTRACE_DIRECT_MULTI	if HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
 	select HAVE_SETUP_PER_CPU_AREA		if PPC64
arch/powerpc/Makefile +2 -0
index a58b1029592c..9fd82c75dcbd 100644
@@ -61,6 +61,8 @@ else
 KBUILD_LDFLAGS_MODULE += $(objtree)/arch/powerpc/lib/crtsavres.o
 endif
 
+KBUILD_RUSTFLAGS += --target=powerpc-unknown-linux-gnu
+
 ifdef CONFIG_CPU_LITTLE_ENDIAN
 KBUILD_CPPFLAGS	+= -mlittle-endian
 KBUILD_LDFLAGS	+= -EL
rust/Makefile +3 -1
index 4dcc2eff51cb..ae22f2c5f0b3 100644
@@ -384,13 +384,15 @@ bindgen_skip_c_flags := -mno-fp-ret-in-387 -mpreferred-stack-boundary=% \
 	-fstrict-flex-arrays=% -fmin-function-alignment=% \
 	-fzero-init-padding-bits=% -mno-fdpic \
 	-fdiagnostics-show-context -fdiagnostics-show-context=% \
-	--param=% --param asan-% -fno-isolate-erroneous-paths-dereference
+	--param=% --param asan-% -fno-isolate-erroneous-paths-dereference \
+	-ffixed-r2 -mmultiple -mno-readonly-in-sdata
 
 # Derived from `scripts/Makefile.clang`.
 BINDGEN_TARGET_x86	:= x86_64-linux-gnu
 BINDGEN_TARGET_arm64	:= aarch64-linux-gnu
 BINDGEN_TARGET_arm	:= arm-linux-gnueabi
 BINDGEN_TARGET_loongarch	:= loongarch64-linux-gnusf
+BINDGEN_TARGET_powerpc	:= powerpc-linux-gnu
 BINDGEN_TARGET_um	:= $(BINDGEN_TARGET_$(SUBARCH))
 BINDGEN_TARGET		:= $(BINDGEN_TARGET_$(SRCARCH))
 
Signature · 1 line
2.53.0
maintainer
#3 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 22 lines · Mukesh Kumar Chaurasiya (IBM)

From: Link Mauve <[email protected]>

For now only Big Endian 32-bit PowerPC is supported, as that is the only hardware I have. This has been tested on the Nintendo Wii so far, but I plan on also using it on the GameCube, Wii U and Apple G4.

These changes aren’t the only ones required to get the kernel to compile and link on PowerPC, libcore will also have to be changed to not use integer division to format u64, u128 and core::time::Duration, otherwise __udivdi3() and __umoddi3() will have to be added. I have tested this change by replacing the three implementations with unimplemented!() and it linked just fine.

arch/powerpc/Makefile +0 -0
index a58b1029592c..9fd82c75dcbd 100644

arch/powerpc/Makefile else · line 64

Quoted 4 lines
arch/powerpc/Makefile +1 -0
@@ -61,6 +61,8 @@ else
 KBUILD_LDFLAGS_MODULE += $(objtree)/arch/powerpc/lib/crtsavres.o
 endif
 
+KBUILD_RUSTFLAGS += --target=powerpc-unknown-linux-gnu

This needs to be a softfloat target.

Alice

#4 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 27 lines · Alice Ryhl
Quoted 25 lines · Mukesh Kumar Chaurasiya (IBM)

From: Link Mauve <[email protected]>

For now only Big Endian 32-bit PowerPC is supported, as that is the only hardware I have. This has been tested on the Nintendo Wii so far, but I plan on also using it on the GameCube, Wii U and Apple G4.

These changes aren’t the only ones required to get the kernel to compile and link on PowerPC, libcore will also have to be changed to not use integer division to format u64, u128 and core::time::Duration, otherwise __udivdi3() and __umoddi3() will have to be added. I have tested this change by replacing the three implementations with unimplemented!() and it linked just fine.

arch/powerpc/Makefile +0 -0
index a58b1029592c..9fd82c75dcbd 100644
@@ -61,6 +61,8 @@ else
 KBUILD_LDFLAGS_MODULE += $(objtree)/arch/powerpc/lib/crtsavres.o
 endif
 
Quoted 3 lines

arch/powerpc/Makefile else · line 64

Quoted 1 line
arch/powerpc/Makefile +1 -0
+KBUILD_RUSTFLAGS += --target=powerpc-unknown-linux-gnu

This needs to be a softfloat target.

Should we come back to describing the target like I did in my first patch[1] in scripts/generate_rust_target.rs, or should I bring that to Rust to create a powerpc-unknown-unknown-softfloat target upstream? Or is there a better third solution I’m not thinking of?

Quoted 3 lines

Alice

[1] https://lore.kernel.org/rust-for-linux/[email protected]/

Signature · 1 line
Link Mauve
#5 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 5 lines · Link Mauve

Should we come back to describing the target like I did in my first patch[1] in scripts/generate_rust_target.rs, or should I bring that to Rust to create a powerpc-unknown-unknown-softfloat target upstream? Or is there a better third solution I’m not thinking of?

We are trying to stop using the custom target specs, so we should ask upstream to give you a built-in target you can use (or equivalently, a flag to do what you need, but I think the idea is to not have such a flag).

i.e. even if you used the custom target JSON, we would still need to ask, since the goal is to remove that script entirely.

Thanks!

Cheers,
Miguel

#6 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 12 lines · Miguel Ojeda
Quoted 5 lines · Link Mauve

Should we come back to describing the target like I did in my first patch[1] in scripts/generate_rust_target.rs, or should I bring that to Rust to create a powerpc-unknown-unknown-softfloat target upstream? Or is there a better third solution I’m not thinking of?

We are trying to stop using the custom target specs, so we should ask upstream to give you a built-in target you can use (or equivalently, a flag to do what you need, but I think the idea is to not have such a flag).

Quoted 7 lines

i.e. even if you used the custom target JSON, we would still need to ask, since the goal is to remove that script entirely.

Thanks!

Cheers,
Miguel

I think disabling altivec, fp and vsx with compiler flag will work.

What is your opinion on this?

Regards,
Mukesh

Quoted 2 lines
#7 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

In reply to Miguel Ojeda
Quoted 12 lines · Miguel Ojeda
Quoted 5 lines · Link Mauve

Should we come back to describing the target like I did in my first patch[1] in scripts/generate_rust_target.rs, or should I bring that to Rust to create a powerpc-unknown-unknown-softfloat target upstream? Or is there a better third solution I’m not thinking of?

We are trying to stop using the custom target specs, so we should ask upstream to give you a built-in target you can use (or equivalently, a flag to do what you need, but I think the idea is to not have such a flag).

Quoted 7 lines

i.e. even if you used the custom target JSON, we would still need to ask, since the goal is to remove that script entirely.

Thanks!

Cheers,
Miguel

Sorry for the spam, my earlier message got rejected.

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

Regards,
Mukesh

maintainer
#8 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 13 lines · Mukesh Kumar Chaurasiya
Quoted 12 lines · Miguel Ojeda
Quoted 5 lines · Link Mauve

Should we come back to describing the target like I did in my first patch[1] in scripts/generate_rust_target.rs, or should I bring that to Rust to create a powerpc-unknown-unknown-softfloat target upstream? Or is there a better third solution I’m not thinking of?

We are trying to stop using the custom target specs, so we should ask upstream to give you a built-in target you can use (or equivalently, a flag to do what you need, but I think the idea is to not have such a flag).

Quoted 6 lines
Quoted 2 lines

i.e. even if you used the custom target JSON, we would still need to ask, since the goal is to remove that script entirely.

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

I think you can and should submit a PR to add a softfloat target to upstream Rust right now, and I believe there should be no issue in accepting that.

If there's a workaround we can use on existing compiler versions without the target, that's great too, but we should get the target in upstream asap.

Alice

#9 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 21 lines · Alice Ryhl
Quoted 13 lines · Mukesh Kumar Chaurasiya
Quoted 12 lines · Miguel Ojeda
Quoted 5 lines · Link Mauve

Should we come back to describing the target like I did in my first patch[1] in scripts/generate_rust_target.rs, or should I bring that to Rust to create a powerpc-unknown-unknown-softfloat target upstream? Or is there a better third solution I’m not thinking of?

We are trying to stop using the custom target specs, so we should ask upstream to give you a built-in target you can use (or equivalently, a flag to do what you need, but I think the idea is to not have such a flag).

Quoted 6 lines
Quoted 2 lines

i.e. even if you used the custom target JSON, we would still need to ask, since the goal is to remove that script entirely.

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

Quoted 7 lines

I think you can and should submit a PR to add a softfloat target to upstream Rust right now, and I believe there should be no issue in accepting that.

If there's a workaround we can use on existing compiler versions without the target, that's great too, but we should get the target in upstream asap.

Cool, sounds good.

Regards,
Mukesh

Quoted 2 lines

Alice

#10 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

In reply to Mukesh Kumar Chaurasiya
Quoted 4 lines · Mukesh Kumar Chaurasiya

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

It is really up to upstream Rust -- for us, i.e. the kernel, it usually doesn't really matter much how things like that are
accomplished: whether via flags, a built-in target, a custom target, etc. However, we need to know what the path to stability is.

My understanding (but I may be wrong) is that upstream Rust prefer we use built-in targets for softfloat instead of disabling via -Ctarget-feature (and that the other options may go away soon and/or will never be stable) -- at least for some cases. For instance, for arm64, please this recent change kernel-side regarding neon as an entry point:

446a8351f160 ("arm64: rust: clean Rust 1.85.0 warning using softfloat target")

So please ask upstream Rust (probably in their Zulip, e.g. in t-compiler or rust-for-linux channels) what you should do for powerpc. They will likely be happy with a PR adding the target (or whatever they decide) as Alice mentions. And until we reach that minimum version (in a year or more), we can use something else meanwhile. But at least we will have a way towards the end goal, if that makes sense.

In case it helps, let me Cc Ralf, Jubilee and Matthew who were involved in some of that discussion in the past, plus the compiler leads.

Cheers,
Miguel

#11 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 12 lines · Miguel Ojeda
Quoted 4 lines · Mukesh Kumar Chaurasiya

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

It is really up to upstream Rust -- for us, i.e. the kernel, it usually doesn't really matter much how things like that are
accomplished: whether via flags, a built-in target, a custom target, etc. However, we need to know what the path to stability is.

Quoted 9 lines

My understanding (but I may be wrong) is that upstream Rust prefer we use built-in targets for softfloat instead of disabling via -Ctarget-feature (and that the other options may go away soon and/or will never be stable) -- at least for some cases. For instance, for arm64, please this recent change kernel-side regarding neon as an entry point:

446a8351f160 ("arm64: rust: clean Rust 1.85.0 warning using softfloat target")

Aah, that makes it clearer.

Quoted 7 lines

So please ask upstream Rust (probably in their Zulip, e.g. in t-compiler or rust-for-linux channels) what you should do for powerpc. They will likely be happy with a PR adding the target (or whatever they decide) as Alice mentions. And until we reach that minimum version (in a year or more), we can use something else meanwhile. But at least we will have a way towards the end goal, if that makes sense.

Yeah makes sense. Will work towards this.

Regards,
Mukesh

Quoted 6 lines

In case it helps, let me Cc Ralf, Jubilee and Matthew who were involved in some of that discussion in the past, plus the compiler leads.

Cheers,
Miguel

#12 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

In reply to Miguel Ojeda

Hi all,

Quoted 28 lines · Miguel Ojeda
Quoted 4 lines · Mukesh Kumar Chaurasiya

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

It is really up to upstream Rust -- for us, i.e. the kernel, it usually doesn't really matter much how things like that are
accomplished: whether via flags, a built-in target, a custom target, etc. However, we need to know what the path to stability is.

My understanding (but I may be wrong) is that upstream Rust prefer we use built-in targets for softfloat instead of disabling via -Ctarget-feature (and that the other options may go away soon and/or will never be stable) -- at least for some cases. For instance, for arm64, please this recent change kernel-side regarding neon as an entry point:

446a8351f160 ("arm64: rust: clean Rust 1.85.0 warning using softfloat target")

So please ask upstream Rust (probably in their Zulip, e.g. in t-compiler or rust-for-linux channels) what you should do for powerpc. They will likely be happy with a PR adding the target (or whatever they decide) as Alice mentions. And until we reach that minimum version (in a year or more), we can use something else meanwhile. But at least we will have a way towards the end goal, if that makes sense.

Quoted 3 lines

In case it helps, let me Cc Ralf, Jubilee and Matthew who were involved in some of that discussion in the past, plus the compiler leads.

Upstream Rust dev here. Indeed we'd strongly prefer if this could use a built-in Rust target; we can work with you on adding a new target if that is needed.
The kernel currently uses a custom JSON target on x86 and that's quite the headache for compiler development: JSON targets are highly unstable and directly expose low-level details of how the compiler internally represents targets. When we change that representation, we update all built-in targets, but of course we cannot update JSON targets. So whenever possible we'd like to move towards reducing the number of JSON targets used by the kernel, not increase it. :)

Kind regards,
Ralf

#13 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 46 lines · Ralf Jung

Hi all,

Quoted 28 lines · Miguel Ojeda
Quoted 4 lines · Mukesh Kumar Chaurasiya

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

It is really up to upstream Rust -- for us, i.e. the kernel, it usually doesn't really matter much how things like that are
accomplished: whether via flags, a built-in target, a custom target, etc. However, we need to know what the path to stability is.

My understanding (but I may be wrong) is that upstream Rust prefer we use built-in targets for softfloat instead of disabling via -Ctarget-feature (and that the other options may go away soon and/or will never be stable) -- at least for some cases. For instance, for arm64, please this recent change kernel-side regarding neon as an entry point:

446a8351f160 ("arm64: rust: clean Rust 1.85.0 warning using softfloat target")

So please ask upstream Rust (probably in their Zulip, e.g. in t-compiler or rust-for-linux channels) what you should do for powerpc. They will likely be happy with a PR adding the target (or whatever they decide) as Alice mentions. And until we reach that minimum version (in a year or more), we can use something else meanwhile. But at least we will have a way towards the end goal, if that makes sense.

Quoted 3 lines

In case it helps, let me Cc Ralf, Jubilee and Matthew who were involved in some of that discussion in the past, plus the compiler leads.

Upstream Rust dev here. Indeed we'd strongly prefer if this could use a built-in Rust target; we can work with you on adding a new target if that is needed.
The kernel currently uses a custom JSON target on x86 and that's quite the headache for compiler development: JSON targets are highly unstable and directly expose low-level details of how the compiler internally represents targets. When we change that representation, we update all built-in targets, but of course we cannot update JSON targets. So whenever possible we'd like to move towards reducing the number of JSON targets used by the kernel, not increase it. :)

Quoted 3 lines

Kind regards,
Ralf

Hey,

Sorry for delayed response. I was out of network zone.

I am not sure about the process of how to get this in rust toolchain. Should I raise an issue of github for this?

Regards,
Mukesh

maintainer
#14 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Quoted 50 lines · Mukesh Kumar Chaurasiya
Quoted 46 lines · Ralf Jung

Hi all,

Quoted 28 lines · Miguel Ojeda
Quoted 4 lines · Mukesh Kumar Chaurasiya

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

It is really up to upstream Rust -- for us, i.e. the kernel, it usually doesn't really matter much how things like that are
accomplished: whether via flags, a built-in target, a custom target, etc. However, we need to know what the path to stability is.

My understanding (but I may be wrong) is that upstream Rust prefer we use built-in targets for softfloat instead of disabling via -Ctarget-feature (and that the other options may go away soon and/or will never be stable) -- at least for some cases. For instance, for arm64, please this recent change kernel-side regarding neon as an entry point:

446a8351f160 ("arm64: rust: clean Rust 1.85.0 warning using softfloat target")

So please ask upstream Rust (probably in their Zulip, e.g. in t-compiler or rust-for-linux channels) what you should do for powerpc. They will likely be happy with a PR adding the target (or whatever they decide) as Alice mentions. And until we reach that minimum version (in a year or more), we can use something else meanwhile. But at least we will have a way towards the end goal, if that makes sense.

Quoted 3 lines

In case it helps, let me Cc Ralf, Jubilee and Matthew who were involved in some of that discussion in the past, plus the compiler leads.

Upstream Rust dev here. Indeed we'd strongly prefer if this could use a built-in Rust target; we can work with you on adding a new target if that is needed.
The kernel currently uses a custom JSON target on x86 and that's quite the headache for compiler development: JSON targets are highly unstable and directly expose low-level details of how the compiler internally represents targets. When we change that representation, we update all built-in targets, but of course we cannot update JSON targets. So whenever possible we'd like to move towards reducing the number of JSON targets used by the kernel, not increase it. :)

Quoted 3 lines

Kind regards,
Ralf

Quoted 6 lines

Hey,

Sorry for delayed response. I was out of network zone.

I am not sure about the process of how to get this in rust toolchain. Should I raise an issue of github for this?

You would need to add a new file to compiler/rustc_target/src/spec/targets in the rustc repository.

If you're not sure what to put there, I would suggest coming up with something that looks plausible, and opening a PR with that. Then others can help you with filling out the target correctly.

Alice

#15 lore raw

Re: [PATCH V6 2/3] rust: Add PowerPC support

Hi all,

Quoted 58 lines · Alice Ryhl
Quoted 50 lines · Mukesh Kumar Chaurasiya
Quoted 46 lines · Ralf Jung

Hi all,

Quoted 28 lines · Miguel Ojeda
Quoted 4 lines · Mukesh Kumar Chaurasiya

I think, disabling altivec, fpu and vsx with compiler flag will work.

What are your opinion on this?

It is really up to upstream Rust -- for us, i.e. the kernel, it usually doesn't really matter much how things like that are
accomplished: whether via flags, a built-in target, a custom target, etc. However, we need to know what the path to stability is.

My understanding (but I may be wrong) is that upstream Rust prefer we use built-in targets for softfloat instead of disabling via -Ctarget-feature (and that the other options may go away soon and/or will never be stable) -- at least for some cases. For instance, for arm64, please this recent change kernel-side regarding neon as an entry point:

    446a8351f160 ("arm64: rust: clean Rust 1.85.0 warning using softfloat target")

So please ask upstream Rust (probably in their Zulip, e.g. in t-compiler or rust-for-linux channels) what you should do for powerpc. They will likely be happy with a PR adding the target (or whatever they decide) as Alice mentions. And until we reach that minimum version (in a year or more), we can use something else meanwhile. But at least we will have a way towards the end goal, if that makes sense.

Quoted 3 lines

In case it helps, let me Cc Ralf, Jubilee and Matthew who were involved in some of that discussion in the past, plus the compiler leads.

Upstream Rust dev here. Indeed we'd strongly prefer if this could use a built-in Rust target; we can work with you on adding a new target if that is needed.
The kernel currently uses a custom JSON target on x86 and that's quite the headache for compiler development: JSON targets are highly unstable and directly expose low-level details of how the compiler internally represents targets. When we change that representation, we update all built-in targets, but of course we cannot update JSON targets. So whenever possible we'd like to move towards reducing the number of JSON targets used by the kernel, not increase it. :)

Quoted 3 lines

Kind regards,
Ralf

Quoted 6 lines

Hey,

Sorry for delayed response. I was out of network zone.

I am not sure about the process of how to get this in rust toolchain. Should I raise an issue of github for this?

Quoted 6 lines

You would need to add a new file to compiler/rustc_target/src/spec/targets in the rustc repository.

If you're not sure what to put there, I would suggest coming up with something that looks plausible, and opening a PR with that. Then others can help you with filling out the target correctly.

Also see the documentation at <https://doc.rust-lang.org/rustc/target-tier-policy.html#adding-a-new-target>.

Kind regards,
Ralf

#16 lore raw

PATCH v6 3/3 powerpc: Enable Rust for ppc64le

In reply to Mukesh Kumar Chaurasiya (IBM)

Enabling rust support for ppc64le only.

Tested on powernv9:

$ uname -rm
6.19.0-rc8+ ppc64le
$ sudo modprobe rust_minimal
[  632.890850] rust_minimal: Rust minimal sample (init)
[  632.890881] rust_minimal: Am I built-in? false
[  632.890898] rust_minimal: test_parameter: 1
$ sudo rmmod rust_minimal
[  648.272832] rust_minimal: My numbers are [72, 108, 200]
[  648.272873] rust_minimal: Rust minimal sample (exit)
$ sudo modprobe rust_print
[  843.410391] rust_print: Rust printing macros sample (init)
[  843.410424] rust_print: Emergency message (level 0) without args
[  843.410451] rust_print: Alert message (level 1) without args
[  843.410477] rust_print: Critical message (level 2) without args
[  843.410503] rust_print: Error message (level 3) without args
[  843.410530] rust_print: Warning message (level 4) without args
[  843.410557] rust_print: Notice message (level 5) without args
[  843.410594] rust_print: Info message (level 6) without args
[  843.410617] rust_print: A line that is continued without args
[  843.410646] rust_print: Emergency message (level 0) with args
[  843.410675] rust_print: Alert message (level 1) with args
[  843.410691] rust_print: Critical message (level 2) with args
[  843.410727] rust_print: Error message (level 3) with args
[  843.410761] rust_print: Warning message (level 4) with args
[  843.410796] rust_print: Notice message (level 5) with args
[  843.410821] rust_print: Info message (level 6) with args
[  843.410854] rust_print: A line that is continued with args
[  843.410892] rust_print: 1
[  843.410895] rust_print: "hello, world"
[  843.410924] rust_print: [samples/rust/rust_print_main.rs:35:5] c = "hello, world"
[  843.410977] rust_print: Arc<dyn Display> says 42
[  843.410979] rust_print: Arc<dyn Display> says hello, world
$ sudo rmmod rust_print
[  843.411003] rust_print: "hello, world"
[  888.499935] rust_print: Rust printing macros sample (exit)
---
 Documentation/rust/arch-support.rst | 2 +-
 arch/powerpc/Kconfig                | 1 +
 arch/powerpc/Makefile               | 7 ++++++-
 rust/Makefile                       | 6 ++++++
 4 files changed, 14 insertions(+), 2 deletions(-)
Documentation/rust/arch-support.rst +1 -1
index 70b9e192a7a0..70bcd726ad0e 100644
@@ -18,7 +18,7 @@ Architecture   Level of support  Constraints
 ``arm``        Maintained        ARMv7 Little Endian only.
 ``arm64``      Maintained        Little Endian only.
 ``loongarch``  Maintained        \-
-``powerpc``    Maintained        32-bit Big Endian only.
+``powerpc``    Maintained        64-bit Little Endian. 32-bit Big Endian.
 ``riscv``      Maintained        ``riscv64`` and LLVM/Clang only.
 ``um``         Maintained        \-
 ``x86``        Maintained        ``x86_64`` only.
arch/powerpc/Kconfig +1 -0
index 17db23b82e91..954af27e10c5 100644
@@ -284,6 +284,7 @@ config PPC
 	select HAVE_RELIABLE_STACKTRACE
 	select HAVE_RSEQ
 	select HAVE_RUST			if PPC32
+	select HAVE_RUST			if PPC64 && CPU_LITTLE_ENDIAN
 	select HAVE_SAMPLE_FTRACE_DIRECT	if HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
 	select HAVE_SAMPLE_FTRACE_DIRECT_MULTI	if HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
 	select HAVE_SETUP_PER_CPU_AREA		if PPC64
arch/powerpc/Makefile +6 -1
index 9fd82c75dcbd..9385db478c59 100644
@@ -61,7 +61,12 @@ else
 KBUILD_LDFLAGS_MODULE += $(objtree)/arch/powerpc/lib/crtsavres.o
 endif
 
-KBUILD_RUSTFLAGS += --target=powerpc-unknown-linux-gnu
+ifdef CONFIG_PPC64
+KBUILD_RUSTFLAGS	+= --target=powerpc64le-unknown-linux-gnu
+KBUILD_RUSTFLAGS	+= -Ctarget-feature=-mma,-vsx,-hard-float,-altivec
+else
+KBUILD_RUSTFLAGS	+= --target=powerpc-unknown-linux-gnu
+endif
 
 ifdef CONFIG_CPU_LITTLE_ENDIAN
 KBUILD_CPPFLAGS	+= -mlittle-endian
rust/Makefile +6 -0
index ae22f2c5f0b3..0aef472c6cf5 100644
@@ -392,7 +392,13 @@ BINDGEN_TARGET_x86	:= x86_64-linux-gnu
 BINDGEN_TARGET_arm64	:= aarch64-linux-gnu
 BINDGEN_TARGET_arm	:= arm-linux-gnueabi
 BINDGEN_TARGET_loongarch	:= loongarch64-linux-gnusf
+
+ifdef CONFIG_PPC64
+BINDGEN_TARGET_powerpc	:= powerpc64le-linux-gnu
+else
 BINDGEN_TARGET_powerpc	:= powerpc-linux-gnu
+endif
+
 BINDGEN_TARGET_um	:= $(BINDGEN_TARGET_$(SUBARCH))
 BINDGEN_TARGET		:= $(BINDGEN_TARGET_$(SRCARCH))
 
Signature · 1 line
2.53.0
#17 lore raw
In reply to Mukesh Kumar Chaurasiya (IBM)
Quoted 11 lines · Mukesh Kumar Chaurasiya (IBM)

Enable experimental rust support for ppc64le and ppc32be. The patch for ppc32 has been provided by Link Mauve[1] and ppc64le support[2] has been merged over it. ppc32 needs some toolchain fixes mentioned in the patch rust: Add PowerPC support and the discussion for that is done here[1].

This has been tested on powernv9 hardware and power10 pseries qemu. I I request Link to test the ppc32 part as i don't have a hardware to test it out.

[1] https://lore.kernel.org/all/[email protected]
[2] https://lore.kernel.org/all/[email protected]

Could see these build issues with the Rust patchset in the compilation of powerpc-next-test This happens when compilation happens only with few threads

# rustc --version
rustc 1.94.0 (4a4ef493e 2026-03-02)

....
  EXPORTS rust/exports_core_generated.h
  BINDGEN rust/bindings/bindings_generated.rs
  BINDGEN rust/bindings/bindings_helpers_generated.rs
  CC      rust/helpers/helpers.o
  EXPORTS rust/exports_helpers_generated.h
  RUSTC L rust/compiler_builtins.o
  RUSTC L rust/ffi.o
  RUSTC PL rust/libproc_macro2.rlib
error[E0464]: multiple candidates for rmeta dependency core found
 --> rust/proc-macro2/marker.rs:4:5
  |
4 | use core::marker::PhantomData;
  |     ^^^^
  |
  = note: candidate #1: /root/.rustup/toolchains/nightly-powerpc64le-unknown-linux-gnu/lib/rustlib/powerpc64le-unknown-linux-gnu/lib/libcore-951759db375eea0c.rmeta = note: candidate #2: ./rust/libcore.rmeta

error[E0119]: conflicting implementations of trait PartialEq for type fallback::Ident --> rust/proc-macro2/fallback.rs:875:1
    |
869 |   impl PartialEq for Ident {
    |   ------------------------ first implementation here
...

875 | / impl<T> PartialEq<T> for Ident
876 | | where
877 | |     T: ?Sized + AsRef<str>,

    | |___________________________^ conflicting implementation for fallback::Ident

error[E0277]: LexError doesn't implement std::fmt::Display
   --> rust/proc-macro2/lib.rs:347:16
    |
347 | impl Error for LexError {}
    |                ^^^^^^^^ unsatisfied trait bound
    |
help: the trait std::fmt::Display is not implemented for LexError
   --> rust/proc-macro2/lib.rs:204:1
    |
204 | pub struct LexError {
    | ^^^^^^^^^^^^^^^^^^^
note: required by a bound in std::error::Error
   --> /root/.rustup/toolchains/nightly-powerpc64le-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/error.rs:59:26 |
 59 | pub trait Error: Debug + Display {
    |                          ^^^^^^^ required by this bound in Error

error[E0277]: LexError doesn't implement Debug
   --> rust/proc-macro2/lib.rs:347:16
    |
347 | impl Error for LexError {}
    |                ^^^^^^^^ unsatisfied trait bound
    |
help: the trait Debug is not implemented for LexError
   --> rust/proc-macro2/lib.rs:204:1
    |
204 | pub struct LexError {
    | ^^^^^^^^^^^^^^^^^^^
    = note: add #[derive(Debug)] to LexError or manually impl Debug for LexError
note: required by a bound in std::error::Error
   --> /root/.rustup/toolchains/nightly-powerpc64le-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/error.rs:59:18 |
 59 | pub trait Error: Debug + Display {
    |                  ^^^^^ required by this bound in Error

error: aborting due to 4 previous errors

......
But when parallelized with more threads (-j 128) compilation passes with out any error There is some ordering of libcore is messing up I guess (I could be wrong) (I have removed the warning of unstable features messages here for cleaner output)

....
  VDSO64SYM include/generated/vdso64-offsets.h
  RUSTC L rust/core.o
  BINDGEN rust/bindings/bindings_generated.rs
  BINDGEN rust/bindings/bindings_helpers_generated.rs
  CC      rust/helpers/helpers.o
  RUSTC PL rust/libproc_macro2.rlib
  BINDGEN rust/uapi/uapi_generated.rs
  RSCPP     rust/kernel/generated_arch_static_branch_asm.rs
  RSCPP     rust/kernel/generated_arch_warn_asm.rs
  RSCPP     rust/kernel/generated_arch_reachable_asm.rs
clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:5:9: warning: '_IOC_SIZEBITS' macro redefined [-Wmacro-redefined] clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:6:9: warning: '_IOC_DIRBITS' macro redefined [-Wmacro-redefined] clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:8:9: warning: '_IOC_NONE' macro redefined [-Wmacro-redefined] clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:10:9: warning: '_IOC_WRITE' macro redefined [-Wmacro-redefined] EXPORTS rust/exports_helpers_generated.h
  RUSTC PL rust/libquote.rlib
  RUSTC PL rust/libsyn.rlib
  RUSTC P rust/libpin_init_internal.so
  RUSTC P rust/libmacros.so
  EXPORTS rust/exports_core_generated.h
  RUSTC L rust/compiler_builtins.o
  RUSTC L rust/ffi.o
  RUSTC L rust/pin_init.o
  RUSTC L rust/build_error.o
  RUSTC L rust/bindings.o
  RUSTC L rust/uapi.o
  EXPORTS rust/exports_bindings_generated.h
  RUSTC L rust/kernel.o
  EXPORTS rust/exports_kernel_generated.h
  LDS     scripts/module.lds
  HOSTCC  usr/gen_init_cpio
  CC      init/main.o
....

Also I see some errors when compiling modules. I am looking at these and any help is welcome. I will pull out Rust patches for now from powerpc-linux next-test branch and once this is restored I will add these patches back to branch for the merge.

Maddy

Quoted 42 lines

Changelog:
V5 -> V6:
- Added a missing Tested by from Venkat which got missed since V3
- Support is marked as Maintained instead of experimental
V5: https://lore.kernel.org/all/[email protected]

V4 -> V5:
- Removed a nested ifdef from PPC64 for Little endian toolchain
V4: https://lore.kernel.org/all/[email protected]

V3 -> V4:
- Co-developed-by header added in patch 1
V3: https://lore.kernel.org/all/[email protected]

V2 -> V3:
- Splited HAVE_RUST in 2 lines
- BINDGEN_TARGET_powerpc initialized before assigning the same to BINDGEN_TARGET
V2: https://lore.kernel.org/all/[email protected]

V1 -> V2:
- jump label fix for rust has been moved to a separate patch
- PPC32 support has been taken
- rust support has been marked experimental
- target.json dependency has been removed
- HAVE_RUST now depends on CPU_LITTLE_ENDIAN for PPC64

Link Mauve (1):
rust: Add PowerPC support

Mukesh Kumar Chaurasiya (IBM) (2):
powerpc/jump_label: adjust inline asm to be consistent
powerpc: Enable Rust for ppc64le

  Documentation/rust/arch-support.rst   |  1 +
  arch/powerpc/Kconfig                  |  2 ++
  arch/powerpc/Makefile                 |  7 +++++++
  arch/powerpc/include/asm/jump_label.h | 23 +++++++++++++----------
  rust/Makefile                         | 10 +++++++++-
  5 files changed, 32 insertions(+), 11 deletions(-)
#18 lore raw
Quoted 136 lines · Madhavan Srinivasan
Quoted 11 lines · Mukesh Kumar Chaurasiya (IBM)

Enable experimental rust support for ppc64le and ppc32be. The patch for ppc32 has been provided by Link Mauve[1] and ppc64le support[2] has been merged over it. ppc32 needs some toolchain fixes mentioned in the patch rust: Add PowerPC support and the discussion for that is done here[1].

This has been tested on powernv9 hardware and power10 pseries qemu. I I request Link to test the ppc32 part as i don't have a hardware to test it out.

[1] https://lore.kernel.org/all/[email protected]
[2] https://lore.kernel.org/all/[email protected]

Could see these build issues with the Rust patchset in the compilation of powerpc-next-test
This happens when compilation happens only with few threads

# rustc --version
rustc 1.94.0 (4a4ef493e 2026-03-02)

....
  EXPORTS rust/exports_core_generated.h
  BINDGEN rust/bindings/bindings_generated.rs
  BINDGEN rust/bindings/bindings_helpers_generated.rs
  CC      rust/helpers/helpers.o
  EXPORTS rust/exports_helpers_generated.h
  RUSTC L rust/compiler_builtins.o
  RUSTC L rust/ffi.o
  RUSTC PL rust/libproc_macro2.rlib
error[E0464]: multiple candidates for rmeta dependency core found
 --> rust/proc-macro2/marker.rs:4:5
  |
4 | use core::marker::PhantomData;
  |     ^^^^
  |
  = note: candidate #1: /root/.rustup/toolchains/nightly-powerpc64le-unknown-linux-gnu/lib/rustlib/powerpc64le-unknown-linux-gnu/lib/libcore-951759db375eea0c.rmeta = note: candidate #2: ./rust/libcore.rmeta

error[E0119]: conflicting implementations of trait PartialEq for type fallback::Ident
   --> rust/proc-macro2/fallback.rs:875:1
    |
869 |   impl PartialEq for Ident {
    |   ------------------------ first implementation here
...

875 | / impl<T> PartialEq<T> for Ident
876 | | where
877 | |     T: ?Sized + AsRef<str>,

    | |___________________________^ conflicting implementation for fallback::Ident

error[E0277]: LexError doesn't implement std::fmt::Display
   --> rust/proc-macro2/lib.rs:347:16
    |
347 | impl Error for LexError {}
    |                ^^^^^^^^ unsatisfied trait bound
    |
help: the trait std::fmt::Display is not implemented for LexError
   --> rust/proc-macro2/lib.rs:204:1
    |
204 | pub struct LexError {
    | ^^^^^^^^^^^^^^^^^^^
note: required by a bound in std::error::Error
   --> /root/.rustup/toolchains/nightly-powerpc64le-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/error.rs:59:26 |
 59 | pub trait Error: Debug + Display {
    |                          ^^^^^^^ required by this bound in Error

error[E0277]: LexError doesn't implement Debug
   --> rust/proc-macro2/lib.rs:347:16
    |
347 | impl Error for LexError {}
    |                ^^^^^^^^ unsatisfied trait bound
    |
help: the trait Debug is not implemented for LexError
   --> rust/proc-macro2/lib.rs:204:1
    |
204 | pub struct LexError {
    | ^^^^^^^^^^^^^^^^^^^
    = note: add #[derive(Debug)] to LexError or manually `impl Debug for
LexError`
note: required by a bound in std::error::Error
   --> /root/.rustup/toolchains/nightly-powerpc64le-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/error.rs:59:18 |
 59 | pub trait Error: Debug + Display {
    |                  ^^^^^ required by this bound in Error

error: aborting due to 4 previous errors

......
But when parallelized with more threads (-j 128) compilation passes with out any error
There is some ordering of libcore is messing up I guess (I could be wrong) (I have removed the warning of unstable features messages here for cleaner output)

....
  VDSO64SYM include/generated/vdso64-offsets.h
  RUSTC L rust/core.o
  BINDGEN rust/bindings/bindings_generated.rs
  BINDGEN rust/bindings/bindings_helpers_generated.rs CC      rust/helpers/helpers.o
  RUSTC PL rust/libproc_macro2.rlib
  BINDGEN rust/uapi/uapi_generated.rs
  RSCPP     rust/kernel/generated_arch_static_branch_asm.rs RSCPP     rust/kernel/generated_arch_warn_asm.rs RSCPP     rust/kernel/generated_arch_reachable_asm.rs clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:5:9: warning:
'_IOC_SIZEBITS' macro redefined [-Wmacro-redefined] clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:6:9: warning:
'_IOC_DIRBITS' macro redefined [-Wmacro-redefined] clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:8:9: warning:
'_IOC_NONE' macro redefined [-Wmacro-redefined]
clang diag: ./arch/powerpc/include/uapi/asm/ioctl.h:10:9: warning:
'_IOC_WRITE' macro redefined [-Wmacro-redefined] EXPORTS rust/exports_helpers_generated.h
  RUSTC PL rust/libquote.rlib
  RUSTC PL rust/libsyn.rlib
  RUSTC P rust/libpin_init_internal.so
  RUSTC P rust/libmacros.so
  EXPORTS rust/exports_core_generated.h
  RUSTC L rust/compiler_builtins.o
  RUSTC L rust/ffi.o
  RUSTC L rust/pin_init.o
  RUSTC L rust/build_error.o
  RUSTC L rust/bindings.o
  RUSTC L rust/uapi.o
  EXPORTS rust/exports_bindings_generated.h
  RUSTC L rust/kernel.o
  EXPORTS rust/exports_kernel_generated.h
  LDS     scripts/module.lds
  HOSTCC  usr/gen_init_cpio
  CC      init/main.o
....

Quoted 8 lines

Also I see some errors when compiling modules. I am looking at these and any help is welcome.
I will pull out Rust patches for now from powerpc-linux next-test branch and once this is
restored I will add these patches back to branch for the merge.

Maddy

Aah this happens because core.o is compiled before libproc_macro2.rlib starts compiling. Once the core.o is compiled it generates the libcore.rmeta, leading to conflict in two libcore.rmeta available.

I can think of 2 solutions here for this,

1. We can make the libproc_macro2.rlib libquote.rlib libsyn.rlib compile befor the core.o is compiled OR
2. We can force these to use the toolchain core's metadata and not look into the kernel's rust directory.

I am still not able to figure out why this is not happening for x86.

What should we do in this case?

Regards,
Mukesh

Quoted 42 lines
Quoted 42 lines

Changelog:
V5 -> V6:
- Added a missing Tested by from Venkat which got missed since V3
- Support is marked as Maintained instead of experimental
V5: https://lore.kernel.org/all/[email protected]

V4 -> V5:
- Removed a nested ifdef from PPC64 for Little endian toolchain
V4: https://lore.kernel.org/all/[email protected]

V3 -> V4:
- Co-developed-by header added in patch 1
V3: https://lore.kernel.org/all/[email protected]

V2 -> V3:
- Splited HAVE_RUST in 2 lines
- BINDGEN_TARGET_powerpc initialized before assigning the same to BINDGEN_TARGET
V2: https://lore.kernel.org/all/[email protected]

V1 -> V2:
- jump label fix for rust has been moved to a separate patch
- PPC32 support has been taken
- rust support has been marked experimental
- target.json dependency has been removed
- HAVE_RUST now depends on CPU_LITTLE_ENDIAN for PPC64

Link Mauve (1):
rust: Add PowerPC support

Mukesh Kumar Chaurasiya (IBM) (2):
powerpc/jump_label: adjust inline asm to be consistent
powerpc: Enable Rust for ppc64le

  Documentation/rust/arch-support.rst   |  1 +
  arch/powerpc/Kconfig                  |  2 ++
  arch/powerpc/Makefile                 |  7 +++++++
  arch/powerpc/include/asm/jump_label.h | 23 +++++++++++++----------
  rust/Makefile                         | 10 +++++++++-
  5 files changed, 32 insertions(+), 11 deletions(-)