public inbox for llvm@lists.linux.dev
 help / color / mirror / Atom feed
* Re: [PATCH v7 1/1] riscv: Allow to downgrade paging mode from the command line
       [not found] ` <20230224100218.1824569-2-alexghiti@rivosinc.com>
@ 2023-02-27 12:31   ` Conor Dooley
  2023-02-27 12:56     ` Ard Biesheuvel
  0 siblings, 1 reply; 4+ messages in thread
From: Conor Dooley @ 2023-02-27 12:31 UTC (permalink / raw)
  To: Alexandre Ghiti
  Cc: Jonathan Corbet, Paul Walmsley, Palmer Dabbelt, Albert Ou,
	Conor Dooley, Ard Biesheuvel, Björn Töpel, linux-doc,
	linux-kernel, linux-riscv, Björn Töpel, nathan, llvm,
	ndesaulniers

[-- Attachment #1: Type: text/plain, Size: 3495 bytes --]

Hey Alex, clang/llvm folk,

On Fri, Feb 24, 2023 at 11:02:18AM +0100, Alexandre Ghiti wrote:
> Add 2 early command line parameters that allow to downgrade satp mode
> (using the same naming as x86):
> - "no5lvl": use a 4-level page table (down from sv57 to sv48)
> - "no4lvl": use a 3-level page table (down from sv57/sv48 to sv39)
> 
> Note that going through the device tree to get the kernel command line
> works with ACPI too since the efi stub creates a device tree anyway with
> the command line.
> 
> In KASAN kernels, we can't use the libfdt that early in the boot process
> since we are not ready to execute instrumented functions. So instead of
> using the "generic" libfdt, we compile our own versions of those functions
> that are not instrumented and that are prefixed so that they do not
> conflict with the generic ones. We also need the non-instrumented versions
> of the string functions and the prefixed versions of memcpy/memmove.
> 
> This is largely inspired by commit aacd149b6238 ("arm64: head: avoid
> relocating the kernel twice for KASLR") from which I removed compilation
> flags that were not relevant to RISC-V at the moment (LTO, SCS, pie).
> 
> Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
> Tested-by: Björn Töpel <bjorn@rivosinc.com>
> Reviewed-by: Björn Töpel <bjorn@rivosinc.com>
> ---
>  .../admin-guide/kernel-parameters.txt         |  5 +-
>  arch/riscv/kernel/Makefile                    |  2 +
>  arch/riscv/kernel/pi/Makefile                 | 37 +++++++++++
>  arch/riscv/kernel/pi/cmdline_early.c          | 62 +++++++++++++++++++

+      4 ld.lld: warning: vmlinux.a(arch/riscv/kernel/pi/string.pi.o):(.init.sdata) is being placed in '.init.sdata'

I'm getting 4 new linker warnings with LLVM=1 clang-15 allmodconfig
builds - but I don't really understand this issue.
Is this spurious, and a just consequence of...

> diff --git a/arch/riscv/kernel/pi/Makefile b/arch/riscv/kernel/pi/Makefile
> new file mode 100644
> index 000000000000..4002ed94b6d3
> --- /dev/null
> +++ b/arch/riscv/kernel/pi/Makefile
> @@ -0,0 +1,37 @@
> +# SPDX-License-Identifier: GPL-2.0
> +# This file was copied from arm64/kernel/pi/Makefile.
> +
> +KBUILD_CFLAGS	:= $(subst $(CC_FLAGS_FTRACE),,$(KBUILD_CFLAGS)) \
> +		   -Os -DDISABLE_BRANCH_PROFILING $(DISABLE_STACKLEAK_PLUGIN) \
> +		   $(call cc-option,-mbranch-protection=none) \
> +		   -I$(srctree)/scripts/dtc/libfdt -fno-stack-protector \
> +		   -D__DISABLE_EXPORTS -ffreestanding \
> +		   -fno-asynchronous-unwind-tables -fno-unwind-tables \
> +		   $(call cc-option,-fno-addrsig)
> +
> +CFLAGS_cmdline_early.o += -D__NO_FORTIFY
> +CFLAGS_lib-fdt_ro.o += -D__NO_FORTIFY
> +
> +GCOV_PROFILE	:= n
> +KASAN_SANITIZE	:= n
> +KCSAN_SANITIZE	:= n
> +UBSAN_SANITIZE	:= n
> +KCOV_INSTRUMENT	:= n
> +
> +$(obj)/%.pi.o: OBJCOPYFLAGS := --prefix-symbols=__pi_ \
> +			       --remove-section=.note.gnu.property \
> +			       --prefix-alloc-sections=.init

...this?

> +$(obj)/%.pi.o: $(obj)/%.o FORCE
> +	$(call if_changed,objcopy)
> +
> +$(obj)/lib-%.o: $(srctree)/lib/%.c FORCE
> +	$(call if_changed_rule,cc_o_c)
> +
> +$(obj)/string.o: $(srctree)/lib/string.c FORCE
> +	$(call if_changed_rule,cc_o_c)
> +
> +$(obj)/ctype.o: $(srctree)/lib/ctype.c FORCE
> +	$(call if_changed_rule,cc_o_c)
> +
> +obj-y		:= cmdline_early.pi.o string.pi.o ctype.pi.o lib-fdt.pi.o lib-fdt_ro.pi.o
> +extra-y		:= $(patsubst %.pi.o,%.o,$(obj-y))

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH v7 1/1] riscv: Allow to downgrade paging mode from the command line
  2023-02-27 12:31   ` [PATCH v7 1/1] riscv: Allow to downgrade paging mode from the command line Conor Dooley
@ 2023-02-27 12:56     ` Ard Biesheuvel
  2023-02-27 13:27       ` Alexandre Ghiti
  0 siblings, 1 reply; 4+ messages in thread
From: Ard Biesheuvel @ 2023-02-27 12:56 UTC (permalink / raw)
  To: Conor Dooley
  Cc: Alexandre Ghiti, Jonathan Corbet, Paul Walmsley, Palmer Dabbelt,
	Albert Ou, Conor Dooley, Björn Töpel, linux-doc,
	linux-kernel, linux-riscv, Björn Töpel, nathan, llvm,
	ndesaulniers

On Mon, 27 Feb 2023 at 13:32, Conor Dooley <conor.dooley@microchip.com> wrote:
>
> Hey Alex, clang/llvm folk,
>
> On Fri, Feb 24, 2023 at 11:02:18AM +0100, Alexandre Ghiti wrote:
> > Add 2 early command line parameters that allow to downgrade satp mode
> > (using the same naming as x86):
> > - "no5lvl": use a 4-level page table (down from sv57 to sv48)
> > - "no4lvl": use a 3-level page table (down from sv57/sv48 to sv39)
> >
> > Note that going through the device tree to get the kernel command line
> > works with ACPI too since the efi stub creates a device tree anyway with
> > the command line.
> >
> > In KASAN kernels, we can't use the libfdt that early in the boot process
> > since we are not ready to execute instrumented functions. So instead of
> > using the "generic" libfdt, we compile our own versions of those functions
> > that are not instrumented and that are prefixed so that they do not
> > conflict with the generic ones. We also need the non-instrumented versions
> > of the string functions and the prefixed versions of memcpy/memmove.
> >
> > This is largely inspired by commit aacd149b6238 ("arm64: head: avoid
> > relocating the kernel twice for KASLR") from which I removed compilation
> > flags that were not relevant to RISC-V at the moment (LTO, SCS, pie).
> >
> > Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
> > Tested-by: Björn Töpel <bjorn@rivosinc.com>
> > Reviewed-by: Björn Töpel <bjorn@rivosinc.com>
> > ---
> >  .../admin-guide/kernel-parameters.txt         |  5 +-
> >  arch/riscv/kernel/Makefile                    |  2 +
> >  arch/riscv/kernel/pi/Makefile                 | 37 +++++++++++
> >  arch/riscv/kernel/pi/cmdline_early.c          | 62 +++++++++++++++++++
>
> +      4 ld.lld: warning: vmlinux.a(arch/riscv/kernel/pi/string.pi.o):(.init.sdata) is being placed in '.init.sdata'
>
> I'm getting 4 new linker warnings with LLVM=1 clang-15 allmodconfig
> builds - but I don't really understand this issue.
> Is this spurious, and a just consequence of...
>
> > diff --git a/arch/riscv/kernel/pi/Makefile b/arch/riscv/kernel/pi/Makefile
> > new file mode 100644
> > index 000000000000..4002ed94b6d3
> > --- /dev/null
> > +++ b/arch/riscv/kernel/pi/Makefile
> > @@ -0,0 +1,37 @@
> > +# SPDX-License-Identifier: GPL-2.0
> > +# This file was copied from arm64/kernel/pi/Makefile.
> > +
> > +KBUILD_CFLAGS        := $(subst $(CC_FLAGS_FTRACE),,$(KBUILD_CFLAGS)) \
> > +                -Os -DDISABLE_BRANCH_PROFILING $(DISABLE_STACKLEAK_PLUGIN) \
> > +                $(call cc-option,-mbranch-protection=none) \
> > +                -I$(srctree)/scripts/dtc/libfdt -fno-stack-protector \
> > +                -D__DISABLE_EXPORTS -ffreestanding \
> > +                -fno-asynchronous-unwind-tables -fno-unwind-tables \
> > +                $(call cc-option,-fno-addrsig)
> > +
> > +CFLAGS_cmdline_early.o += -D__NO_FORTIFY
> > +CFLAGS_lib-fdt_ro.o += -D__NO_FORTIFY
> > +
> > +GCOV_PROFILE := n
> > +KASAN_SANITIZE       := n
> > +KCSAN_SANITIZE       := n
> > +UBSAN_SANITIZE       := n
> > +KCOV_INSTRUMENT      := n
> > +
> > +$(obj)/%.pi.o: OBJCOPYFLAGS := --prefix-symbols=__pi_ \
> > +                            --remove-section=.note.gnu.property \
> > +                            --prefix-alloc-sections=.init
>
> ...this?
>

Yes. if the input object has a .sdata section, the output .pi.o object
will have a .init.sdata section, and if the common linker script does
not place it explicitly, you will get a warning. Note that we switched
to --orphan-handling=error on other arches, as the linker sometimes
does unexpected things with sections it doesn't know about.

I'd imagine this needs to be combined with .init.data

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH v7 1/1] riscv: Allow to downgrade paging mode from the command line
  2023-02-27 12:56     ` Ard Biesheuvel
@ 2023-02-27 13:27       ` Alexandre Ghiti
  2023-02-27 13:42         ` Conor Dooley
  0 siblings, 1 reply; 4+ messages in thread
From: Alexandre Ghiti @ 2023-02-27 13:27 UTC (permalink / raw)
  To: Ard Biesheuvel, Conor Dooley
  Cc: Alexandre Ghiti, Jonathan Corbet, Paul Walmsley, Palmer Dabbelt,
	Albert Ou, Conor Dooley, Björn Töpel, linux-doc,
	linux-kernel, linux-riscv, Björn Töpel, nathan, llvm,
	ndesaulniers


On 2/27/23 13:56, Ard Biesheuvel wrote:
> On Mon, 27 Feb 2023 at 13:32, Conor Dooley <conor.dooley@microchip.com> wrote:
>> Hey Alex, clang/llvm folk,
>>
>> On Fri, Feb 24, 2023 at 11:02:18AM +0100, Alexandre Ghiti wrote:
>>> Add 2 early command line parameters that allow to downgrade satp mode
>>> (using the same naming as x86):
>>> - "no5lvl": use a 4-level page table (down from sv57 to sv48)
>>> - "no4lvl": use a 3-level page table (down from sv57/sv48 to sv39)
>>>
>>> Note that going through the device tree to get the kernel command line
>>> works with ACPI too since the efi stub creates a device tree anyway with
>>> the command line.
>>>
>>> In KASAN kernels, we can't use the libfdt that early in the boot process
>>> since we are not ready to execute instrumented functions. So instead of
>>> using the "generic" libfdt, we compile our own versions of those functions
>>> that are not instrumented and that are prefixed so that they do not
>>> conflict with the generic ones. We also need the non-instrumented versions
>>> of the string functions and the prefixed versions of memcpy/memmove.
>>>
>>> This is largely inspired by commit aacd149b6238 ("arm64: head: avoid
>>> relocating the kernel twice for KASLR") from which I removed compilation
>>> flags that were not relevant to RISC-V at the moment (LTO, SCS, pie).
>>>
>>> Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
>>> Tested-by: Björn Töpel <bjorn@rivosinc.com>
>>> Reviewed-by: Björn Töpel <bjorn@rivosinc.com>
>>> ---
>>>   .../admin-guide/kernel-parameters.txt         |  5 +-
>>>   arch/riscv/kernel/Makefile                    |  2 +
>>>   arch/riscv/kernel/pi/Makefile                 | 37 +++++++++++
>>>   arch/riscv/kernel/pi/cmdline_early.c          | 62 +++++++++++++++++++
>> +      4 ld.lld: warning: vmlinux.a(arch/riscv/kernel/pi/string.pi.o):(.init.sdata) is being placed in '.init.sdata'
>>
>> I'm getting 4 new linker warnings with LLVM=1 clang-15 allmodconfig
>> builds - but I don't really understand this issue.
>> Is this spurious, and a just consequence of...
>>
>>> diff --git a/arch/riscv/kernel/pi/Makefile b/arch/riscv/kernel/pi/Makefile
>>> new file mode 100644
>>> index 000000000000..4002ed94b6d3
>>> --- /dev/null
>>> +++ b/arch/riscv/kernel/pi/Makefile
>>> @@ -0,0 +1,37 @@
>>> +# SPDX-License-Identifier: GPL-2.0
>>> +# This file was copied from arm64/kernel/pi/Makefile.
>>> +
>>> +KBUILD_CFLAGS        := $(subst $(CC_FLAGS_FTRACE),,$(KBUILD_CFLAGS)) \
>>> +                -Os -DDISABLE_BRANCH_PROFILING $(DISABLE_STACKLEAK_PLUGIN) \
>>> +                $(call cc-option,-mbranch-protection=none) \
>>> +                -I$(srctree)/scripts/dtc/libfdt -fno-stack-protector \
>>> +                -D__DISABLE_EXPORTS -ffreestanding \
>>> +                -fno-asynchronous-unwind-tables -fno-unwind-tables \
>>> +                $(call cc-option,-fno-addrsig)
>>> +
>>> +CFLAGS_cmdline_early.o += -D__NO_FORTIFY
>>> +CFLAGS_lib-fdt_ro.o += -D__NO_FORTIFY
>>> +
>>> +GCOV_PROFILE := n
>>> +KASAN_SANITIZE       := n
>>> +KCSAN_SANITIZE       := n
>>> +UBSAN_SANITIZE       := n
>>> +KCOV_INSTRUMENT      := n
>>> +
>>> +$(obj)/%.pi.o: OBJCOPYFLAGS := --prefix-symbols=__pi_ \
>>> +                            --remove-section=.note.gnu.property \
>>> +                            --prefix-alloc-sections=.init
>> ...this?
>>
> Yes. if the input object has a .sdata section, the output .pi.o object
> will have a .init.sdata section, and if the common linker script does
> not place it explicitly, you will get a warning. Note that we switched
> to --orphan-handling=error on other arches, as the linker sometimes


It seems orphan-handling is set to "error" only when WERROR is set (see 
CONFIG_LD_ORPHAN_WARN_LEVEL).


> does unexpected things with sections it doesn't know about.
>
> I'd imagine this needs to be combined with .init.data


Yes, gcc does not produce this section but others that I placed in a new 
section called .init.pidata: I did not want to expand INIT_DATA_SECTION 
macro because we won't need that anymore when we implement all the 
string functions.

So I'll add this new .init.sdata section to .init.pidata in the v8!

Thanks,


>
> _______________________________________________
> linux-riscv mailing list
> linux-riscv@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH v7 1/1] riscv: Allow to downgrade paging mode from the command line
  2023-02-27 13:27       ` Alexandre Ghiti
@ 2023-02-27 13:42         ` Conor Dooley
  0 siblings, 0 replies; 4+ messages in thread
From: Conor Dooley @ 2023-02-27 13:42 UTC (permalink / raw)
  To: Alexandre Ghiti
  Cc: Ard Biesheuvel, Alexandre Ghiti, Jonathan Corbet, Paul Walmsley,
	Palmer Dabbelt, Albert Ou, Conor Dooley, Björn Töpel,
	linux-doc, linux-kernel, linux-riscv, Björn Töpel,
	nathan, llvm, ndesaulniers

[-- Attachment #1: Type: text/plain, Size: 492 bytes --]

On Mon, Feb 27, 2023 at 02:27:11PM +0100, Alexandre Ghiti wrote:
> On 2/27/23 13:56, Ard Biesheuvel wrote:

> > Note that we switched
> > to --orphan-handling=error on other arches, as the linker sometimes

> It seems orphan-handling is set to "error" only when WERROR is set (see
> CONFIG_LD_ORPHAN_WARN_LEVEL).

My CI explicitly disables CONFIG_WERROR for allmodconfig builds so that
it can run with W=1, it is indeed set to "error" for regular
allmodconfig runs.

Thanks Alex/Ard,
Conor.


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2023-02-27 13:43 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <20230224100218.1824569-1-alexghiti@rivosinc.com>
     [not found] ` <20230224100218.1824569-2-alexghiti@rivosinc.com>
2023-02-27 12:31   ` [PATCH v7 1/1] riscv: Allow to downgrade paging mode from the command line Conor Dooley
2023-02-27 12:56     ` Ard Biesheuvel
2023-02-27 13:27       ` Alexandre Ghiti
2023-02-27 13:42         ` Conor Dooley

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox