| Message ID | 20231116130331.241395-1-wangrui@loongson.cn (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | LoongArch: Record pc instead of offset in la-abs relocation | expand |
Hi Rui, On Thu, Nov 16, 2023 at 09:03:31PM +0800, WANG Rui wrote: > To clarify, the previous version functioned flawlessly. However, it's > worth noting that the LLVM's LoongArch backend currently lacks support > for cross-section label calculations. With this patch, we enable the use > of clang to compile relocatable kernels. > > Signed-off-by: WANG Rui <wangrui@loongson.cn> Thanks a lot for the patch! This fixes the CONFIG_RELOCATABLE build for me as well. Tested-by: Nathan Chancellor <nathan@kernel.org> Something I noticed while testing is that a kernel linked with ld.lld does not boot while one linked with ld.bfd did. I think this might be the same issue that Xuerui filed on our issue tracker but I figured I would mention it in case not: https://github.com/ClangBuiltLinux/linux/issues/1883 Cheers, Nathan > --- > arch/loongarch/include/asm/asmmacro.h | 3 +-- > arch/loongarch/include/asm/setup.h | 2 +- > arch/loongarch/kernel/relocate.c | 2 +- > 3 files changed, 3 insertions(+), 4 deletions(-) > > diff --git a/arch/loongarch/include/asm/asmmacro.h b/arch/loongarch/include/asm/asmmacro.h > index c9544f358c33..655db7d7a427 100644 > --- a/arch/loongarch/include/asm/asmmacro.h > +++ b/arch/loongarch/include/asm/asmmacro.h > @@ -609,8 +609,7 @@ > lu32i.d \reg, 0 > lu52i.d \reg, \reg, 0 > .pushsection ".la_abs", "aw", %progbits > - 768: > - .dword 768b-766b > + .dword 766b > .dword \sym > .popsection > #endif > diff --git a/arch/loongarch/include/asm/setup.h b/arch/loongarch/include/asm/setup.h > index a0bc159ce8bd..ee52fb1e9963 100644 > --- a/arch/loongarch/include/asm/setup.h > +++ b/arch/loongarch/include/asm/setup.h > @@ -25,7 +25,7 @@ extern void set_merr_handler(unsigned long offset, void *addr, unsigned long len > #ifdef CONFIG_RELOCATABLE > > struct rela_la_abs { > - long offset; > + long pc; > long symvalue; > }; > > diff --git a/arch/loongarch/kernel/relocate.c b/arch/loongarch/kernel/relocate.c > index 6c3eff9af9fb..288b739ca88d 100644 > --- a/arch/loongarch/kernel/relocate.c > +++ b/arch/loongarch/kernel/relocate.c > @@ -52,7 +52,7 @@ static inline void __init relocate_absolute(long random_offset) > for (p = begin; (void *)p < end; p++) { > long v = p->symvalue; > uint32_t lu12iw, ori, lu32id, lu52id; > - union loongarch_instruction *insn = (void *)p - p->offset; > + union loongarch_instruction *insn = (void *)p->pc; > > lu12iw = (v >> 12) & 0xfffff; > ori = v & 0xfff; > -- > 2.42.1 > >
Applied to loongarch-fixes, thanks. Huacai On Tue, Nov 21, 2023 at 7:08 AM Nathan Chancellor <nathan@kernel.org> wrote: > > Hi Rui, > > On Thu, Nov 16, 2023 at 09:03:31PM +0800, WANG Rui wrote: > > To clarify, the previous version functioned flawlessly. However, it's > > worth noting that the LLVM's LoongArch backend currently lacks support > > for cross-section label calculations. With this patch, we enable the use > > of clang to compile relocatable kernels. > > > > Signed-off-by: WANG Rui <wangrui@loongson.cn> > > Thanks a lot for the patch! This fixes the CONFIG_RELOCATABLE build for > me as well. > > Tested-by: Nathan Chancellor <nathan@kernel.org> > > Something I noticed while testing is that a kernel linked with ld.lld > does not boot while one linked with ld.bfd did. I think this might be > the same issue that Xuerui filed on our issue tracker but I figured I > would mention it in case not: > https://github.com/ClangBuiltLinux/linux/issues/1883 > > Cheers, > Nathan > > > --- > > arch/loongarch/include/asm/asmmacro.h | 3 +-- > > arch/loongarch/include/asm/setup.h | 2 +- > > arch/loongarch/kernel/relocate.c | 2 +- > > 3 files changed, 3 insertions(+), 4 deletions(-) > > > > diff --git a/arch/loongarch/include/asm/asmmacro.h b/arch/loongarch/include/asm/asmmacro.h > > index c9544f358c33..655db7d7a427 100644 > > --- a/arch/loongarch/include/asm/asmmacro.h > > +++ b/arch/loongarch/include/asm/asmmacro.h > > @@ -609,8 +609,7 @@ > > lu32i.d \reg, 0 > > lu52i.d \reg, \reg, 0 > > .pushsection ".la_abs", "aw", %progbits > > - 768: > > - .dword 768b-766b > > + .dword 766b > > .dword \sym > > .popsection > > #endif > > diff --git a/arch/loongarch/include/asm/setup.h b/arch/loongarch/include/asm/setup.h > > index a0bc159ce8bd..ee52fb1e9963 100644 > > --- a/arch/loongarch/include/asm/setup.h > > +++ b/arch/loongarch/include/asm/setup.h > > @@ -25,7 +25,7 @@ extern void set_merr_handler(unsigned long offset, void *addr, unsigned long len > > #ifdef CONFIG_RELOCATABLE > > > > struct rela_la_abs { > > - long offset; > > + long pc; > > long symvalue; > > }; > > > > diff --git a/arch/loongarch/kernel/relocate.c b/arch/loongarch/kernel/relocate.c > > index 6c3eff9af9fb..288b739ca88d 100644 > > --- a/arch/loongarch/kernel/relocate.c > > +++ b/arch/loongarch/kernel/relocate.c > > @@ -52,7 +52,7 @@ static inline void __init relocate_absolute(long random_offset) > > for (p = begin; (void *)p < end; p++) { > > long v = p->symvalue; > > uint32_t lu12iw, ori, lu32id, lu52id; > > - union loongarch_instruction *insn = (void *)p - p->offset; > > + union loongarch_instruction *insn = (void *)p->pc; > > > > lu12iw = (v >> 12) & 0xfffff; > > ori = v & 0xfff; > > -- > > 2.42.1 > > > >
diff --git a/arch/loongarch/include/asm/asmmacro.h b/arch/loongarch/include/asm/asmmacro.h index c9544f358c33..655db7d7a427 100644 --- a/arch/loongarch/include/asm/asmmacro.h +++ b/arch/loongarch/include/asm/asmmacro.h @@ -609,8 +609,7 @@ lu32i.d \reg, 0 lu52i.d \reg, \reg, 0 .pushsection ".la_abs", "aw", %progbits - 768: - .dword 768b-766b + .dword 766b .dword \sym .popsection #endif diff --git a/arch/loongarch/include/asm/setup.h b/arch/loongarch/include/asm/setup.h index a0bc159ce8bd..ee52fb1e9963 100644 --- a/arch/loongarch/include/asm/setup.h +++ b/arch/loongarch/include/asm/setup.h @@ -25,7 +25,7 @@ extern void set_merr_handler(unsigned long offset, void *addr, unsigned long len #ifdef CONFIG_RELOCATABLE struct rela_la_abs { - long offset; + long pc; long symvalue; }; diff --git a/arch/loongarch/kernel/relocate.c b/arch/loongarch/kernel/relocate.c index 6c3eff9af9fb..288b739ca88d 100644 --- a/arch/loongarch/kernel/relocate.c +++ b/arch/loongarch/kernel/relocate.c @@ -52,7 +52,7 @@ static inline void __init relocate_absolute(long random_offset) for (p = begin; (void *)p < end; p++) { long v = p->symvalue; uint32_t lu12iw, ori, lu32id, lu52id; - union loongarch_instruction *insn = (void *)p - p->offset; + union loongarch_instruction *insn = (void *)p->pc; lu12iw = (v >> 12) & 0xfffff; ori = v & 0xfff;
To clarify, the previous version functioned flawlessly. However, it's worth noting that the LLVM's LoongArch backend currently lacks support for cross-section label calculations. With this patch, we enable the use of clang to compile relocatable kernels. Signed-off-by: WANG Rui <wangrui@loongson.cn> --- arch/loongarch/include/asm/asmmacro.h | 3 +-- arch/loongarch/include/asm/setup.h | 2 +- arch/loongarch/kernel/relocate.c | 2 +- 3 files changed, 3 insertions(+), 4 deletions(-)