diff mbox series

MIPS: Loongson2ef: Remove unnecessary {as,cc}-option calls

Message ID 20211207175951.135400-1-nathan@kernel.org (mailing list archive)
State Accepted
Commit 13ceb48bc19c563e05f407e663aaa4105fdfd408
Headers show
Series MIPS: Loongson2ef: Remove unnecessary {as,cc}-option calls | expand

Commit Message

Nathan Chancellor Dec. 7, 2021, 5:59 p.m. UTC
When building with LLVM's integrated assembler, the build errors because
it does not implement -mfix-loongson2f-{jump,nop}:

arch/mips/loongson2ef/Platform:36: *** only binutils >= 2.20.2 have needed option -mfix-loongson2f-nop.  Stop.

The error is a little misleading because binutils are not being used in
this case.

To clear this up, remove the as-option calls because binutils 2.23 is
the minimum supported version for building the kernel. At the same time,
remove the cc-option calls for the '-march=' flags, as GCC 5.1.0 is the
minimum supported version.

This change will not fix the LLVM build for CONFIG_CPU_LOONGSON2{E,F},
as it does not implement the loongson2{e,f} march arguments (nor r4600,
so it will error prior to this change) nor the assembler flags mentioned
above but it will make the errors more obvious.

Link: https://github.com/ClangBuiltLinux/linux/issues/1529
Reported-by: Ryutaroh Matsumoto <ryutaroh@ict.e.titech.ac.jp>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---

This dependency on certain toolchain flags should probably be moved into
Kconfig at some point so that users cannot select configurations that
they do not have support for so that their builds do not error. However,
from a brief survey, there is not a clean way to codify these
dependencies at the moment because the CPU configs are selected by the
individual machines that implement them, meaning that the dependencies
would need to be added to all the machine configs (as 'select ...'
overrides 'depends on ...'), which is outside the scope of this patch.
Furthermore, one could argue that it is better for the user to get a big
error when they are missing support for something, rather than the
configs getting disabled silently, especially if they are critical to
the machine.

 arch/mips/loongson2ef/Platform | 19 ++++---------------
 1 file changed, 4 insertions(+), 15 deletions(-)


base-commit: 0fcfb00b28c0b7884635dacf38e46d60bf3d4eb1

Comments

Thomas Bogendoerfer Dec. 9, 2021, 9:30 a.m. UTC | #1
On Tue, Dec 07, 2021 at 10:59:51AM -0700, Nathan Chancellor wrote:
> When building with LLVM's integrated assembler, the build errors because
> it does not implement -mfix-loongson2f-{jump,nop}:
> 
> arch/mips/loongson2ef/Platform:36: *** only binutils >= 2.20.2 have needed option -mfix-loongson2f-nop.  Stop.
> 
> The error is a little misleading because binutils are not being used in
> this case.
> 
> To clear this up, remove the as-option calls because binutils 2.23 is
> the minimum supported version for building the kernel. At the same time,
> remove the cc-option calls for the '-march=' flags, as GCC 5.1.0 is the
> minimum supported version.
> 
> This change will not fix the LLVM build for CONFIG_CPU_LOONGSON2{E,F},
> as it does not implement the loongson2{e,f} march arguments (nor r4600,
> so it will error prior to this change) nor the assembler flags mentioned
> above but it will make the errors more obvious.
> 
> Link: https://github.com/ClangBuiltLinux/linux/issues/1529
> Reported-by: Ryutaroh Matsumoto <ryutaroh@ict.e.titech.ac.jp>
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
> 
> This dependency on certain toolchain flags should probably be moved into
> Kconfig at some point so that users cannot select configurations that
> they do not have support for so that their builds do not error. However,
> from a brief survey, there is not a clean way to codify these
> dependencies at the moment because the CPU configs are selected by the
> individual machines that implement them, meaning that the dependencies
> would need to be added to all the machine configs (as 'select ...'
> overrides 'depends on ...'), which is outside the scope of this patch.
> Furthermore, one could argue that it is better for the user to get a big
> error when they are missing support for something, rather than the
> configs getting disabled silently, especially if they are critical to
> the machine.
> 
>  arch/mips/loongson2ef/Platform | 19 ++++---------------
>  1 file changed, 4 insertions(+), 15 deletions(-)

applied to mips-next.

Thomas.
diff mbox series

Patch

diff --git a/arch/mips/loongson2ef/Platform b/arch/mips/loongson2ef/Platform
index ae023b9a1c51..50e659aca543 100644
--- a/arch/mips/loongson2ef/Platform
+++ b/arch/mips/loongson2ef/Platform
@@ -2,12 +2,9 @@ 
 # Loongson Processors' Support
 #
 
-# Only gcc >= 4.4 have Loongson specific support
 cflags-$(CONFIG_CPU_LOONGSON2EF)	+= -Wa,--trap
-cflags-$(CONFIG_CPU_LOONGSON2E) += \
-	$(call cc-option,-march=loongson2e,-march=r4600)
-cflags-$(CONFIG_CPU_LOONGSON2F) += \
-	$(call cc-option,-march=loongson2f,-march=r4600)
+cflags-$(CONFIG_CPU_LOONGSON2E) += -march=loongson2e
+cflags-$(CONFIG_CPU_LOONGSON2F) += -march=loongson2f
 #
 # Some versions of binutils, not currently mainline as of 2019/02/04, support
 # an -mfix-loongson3-llsc flag which emits a sync prior to each ll instruction
@@ -32,16 +29,8 @@  cflags-$(CONFIG_CPU_LOONGSON2EF)	+= $(call as-option,-Wa$(comma)-mno-fix-loongso
 
 # Enable the workarounds for Loongson2f
 ifdef CONFIG_CPU_LOONGSON2F_WORKAROUNDS
-  ifeq ($(call as-option,-Wa$(comma)-mfix-loongson2f-nop,),)
-    $(error only binutils >= 2.20.2 have needed option -mfix-loongson2f-nop)
-  else
-    cflags-$(CONFIG_CPU_NOP_WORKAROUNDS) += -Wa$(comma)-mfix-loongson2f-nop
-  endif
-  ifeq ($(call as-option,-Wa$(comma)-mfix-loongson2f-jump,),)
-    $(error only binutils >= 2.20.2 have needed option -mfix-loongson2f-jump)
-  else
-    cflags-$(CONFIG_CPU_JUMP_WORKAROUNDS) += -Wa$(comma)-mfix-loongson2f-jump
-  endif
+cflags-$(CONFIG_CPU_NOP_WORKAROUNDS) += -Wa,-mfix-loongson2f-nop
+cflags-$(CONFIG_CPU_JUMP_WORKAROUNDS) += -Wa,-mfix-loongson2f-jump
 endif
 
 # Some -march= flags enable MMI instructions, and GCC complains about that