* [PATCH 0/2] Fix some issue on RV32
@ 2018-11-12 4:07 Zong Li
2018-11-12 4:07 ` Zong Li
` (2 more replies)
0 siblings, 3 replies; 12+ messages in thread
From: Zong Li @ 2018-11-12 4:07 UTC (permalink / raw)
To: linux-riscv
This patches fix the build fail and kernel module install failed on RV32.
Zong Li (2):
RISC-V: Request stat64 on RV32
RISC-V: Support MODULE_SECTIONS mechanism on RV32
arch/riscv/Kconfig | 2 +-
arch/riscv/include/asm/module.h | 28 +++++++++++++++-------------
arch/riscv/include/asm/unistd.h | 1 +
arch/riscv/kernel/module-sections.c | 30 ++++++++++++++++--------------
4 files changed, 33 insertions(+), 28 deletions(-)
--
2.7.4
^ permalink raw reply [flat|nested] 12+ messages in thread* [PATCH 0/2] Fix some issue on RV32 2018-11-12 4:07 [PATCH 0/2] Fix some issue on RV32 Zong Li @ 2018-11-12 4:07 ` Zong Li 2018-11-12 4:07 ` [PATCH 1/2] RISC-V: Request stat64 " Zong Li 2018-11-12 4:07 ` [PATCH 2/2] RISC-V: Support MODULE_SECTIONS mechanism " Zong Li 2 siblings, 0 replies; 12+ messages in thread From: Zong Li @ 2018-11-12 4:07 UTC (permalink / raw) To: palmer, aou, linux-riscv, linux-kernel, zong; +Cc: Zong Li This patches fix the build fail and kernel module install failed on RV32. Zong Li (2): RISC-V: Request stat64 on RV32 RISC-V: Support MODULE_SECTIONS mechanism on RV32 arch/riscv/Kconfig | 2 +- arch/riscv/include/asm/module.h | 28 +++++++++++++++------------- arch/riscv/include/asm/unistd.h | 1 + arch/riscv/kernel/module-sections.c | 30 ++++++++++++++++-------------- 4 files changed, 33 insertions(+), 28 deletions(-) -- 2.7.4 _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv ^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 4:07 [PATCH 0/2] Fix some issue on RV32 Zong Li 2018-11-12 4:07 ` Zong Li @ 2018-11-12 4:07 ` Zong Li 2018-11-12 4:07 ` Zong Li 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 4:07 ` [PATCH 2/2] RISC-V: Support MODULE_SECTIONS mechanism " Zong Li 2 siblings, 2 replies; 12+ messages in thread From: Zong Li @ 2018-11-12 4:07 UTC (permalink / raw) To: linux-riscv The stat64 family that is used on 32-bit architectures to replace newstat. Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' Signed-off-by: Zong Li <zong@andestech.com> --- arch/riscv/include/asm/unistd.h | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h index eff7aa9..a4aade9 100644 --- a/arch/riscv/include/asm/unistd.h +++ b/arch/riscv/include/asm/unistd.h @@ -18,5 +18,6 @@ #define __ARCH_WANT_NEW_STAT #define __ARCH_WANT_SYS_CLONE +#define __ARCH_WANT_STAT64 #include <uapi/asm/unistd.h> #include <uapi/asm/syscalls.h> -- 2.7.4 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 4:07 ` [PATCH 1/2] RISC-V: Request stat64 " Zong Li @ 2018-11-12 4:07 ` Zong Li 2018-11-12 6:19 ` David Abdurachmanov 1 sibling, 0 replies; 12+ messages in thread From: Zong Li @ 2018-11-12 4:07 UTC (permalink / raw) To: palmer, aou, linux-riscv, linux-kernel, zong; +Cc: Zong Li The stat64 family that is used on 32-bit architectures to replace newstat. Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' Signed-off-by: Zong Li <zong@andestech.com> --- arch/riscv/include/asm/unistd.h | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h index eff7aa9..a4aade9 100644 --- a/arch/riscv/include/asm/unistd.h +++ b/arch/riscv/include/asm/unistd.h @@ -18,5 +18,6 @@ #define __ARCH_WANT_NEW_STAT #define __ARCH_WANT_SYS_CLONE +#define __ARCH_WANT_STAT64 #include <uapi/asm/unistd.h> #include <uapi/asm/syscalls.h> -- 2.7.4 _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 4:07 ` [PATCH 1/2] RISC-V: Request stat64 " Zong Li 2018-11-12 4:07 ` Zong Li @ 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 6:19 ` David Abdurachmanov ` (2 more replies) 1 sibling, 3 replies; 12+ messages in thread From: David Abdurachmanov @ 2018-11-12 6:19 UTC (permalink / raw) To: linux-riscv On Mon, Nov 12, 2018 at 5:10 AM Zong Li <zongbox@gmail.com> wrote: > > The stat64 family that is used on 32-bit architectures to replace > newstat. > > Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request > newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' > > Signed-off-by: Zong Li <zong@andestech.com> > --- > arch/riscv/include/asm/unistd.h | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h > index eff7aa9..a4aade9 100644 > --- a/arch/riscv/include/asm/unistd.h > +++ b/arch/riscv/include/asm/unistd.h > @@ -18,5 +18,6 @@ > > #define __ARCH_WANT_NEW_STAT > #define __ARCH_WANT_SYS_CLONE > +#define __ARCH_WANT_STAT64 > #include <uapi/asm/unistd.h> > #include <uapi/asm/syscalls.h> > -- See: http://lists.infradead.org/pipermail/linux-riscv/2018-November/002087.html The plan is not to have old stat syscalls and support statx on riscv32, which is y2038 safe. The issue you see is a bug in include/uapi/asm-generic/unistd.h. Marcin (CC) already sent a patch to Arnd (CC) IIRC. Basically without __ARCH_WANT_NEW_STAT or __ARCH_WANT_STAT64 two macros are not defined: __NR3264_fstatat and __NR3264_fstat. Which is later used (without any guards): 763 #define __NR_newfstatat __NR3264_fstatat 764 #define __NR_fstat __NR3264_fstat ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 6:19 ` David Abdurachmanov @ 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 8:29 ` Zong Li 2018-11-12 17:01 ` Palmer Dabbelt 2 siblings, 0 replies; 12+ messages in thread From: David Abdurachmanov @ 2018-11-12 6:19 UTC (permalink / raw) To: zongbox Cc: Marcin Juszkiewicz, aou, Arnd Bergmann, Palmer Dabbelt, linux-kernel, zong, linux-riscv On Mon, Nov 12, 2018 at 5:10 AM Zong Li <zongbox@gmail.com> wrote: > > The stat64 family that is used on 32-bit architectures to replace > newstat. > > Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request > newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' > > Signed-off-by: Zong Li <zong@andestech.com> > --- > arch/riscv/include/asm/unistd.h | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h > index eff7aa9..a4aade9 100644 > --- a/arch/riscv/include/asm/unistd.h > +++ b/arch/riscv/include/asm/unistd.h > @@ -18,5 +18,6 @@ > > #define __ARCH_WANT_NEW_STAT > #define __ARCH_WANT_SYS_CLONE > +#define __ARCH_WANT_STAT64 > #include <uapi/asm/unistd.h> > #include <uapi/asm/syscalls.h> > -- See: http://lists.infradead.org/pipermail/linux-riscv/2018-November/002087.html The plan is not to have old stat syscalls and support statx on riscv32, which is y2038 safe. The issue you see is a bug in include/uapi/asm-generic/unistd.h. Marcin (CC) already sent a patch to Arnd (CC) IIRC. Basically without __ARCH_WANT_NEW_STAT or __ARCH_WANT_STAT64 two macros are not defined: __NR3264_fstatat and __NR3264_fstat. Which is later used (without any guards): 763 #define __NR_newfstatat __NR3264_fstatat 764 #define __NR_fstat __NR3264_fstat _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv ^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 6:19 ` David Abdurachmanov @ 2018-11-12 8:29 ` Zong Li 2018-11-12 8:29 ` Zong Li 2018-11-12 17:01 ` Palmer Dabbelt 2 siblings, 1 reply; 12+ messages in thread From: Zong Li @ 2018-11-12 8:29 UTC (permalink / raw) To: linux-riscv David Abdurachmanov <david.abdurachmanov@gmail.com> ? 2018?11?12? ?? ??2:19??? > > On Mon, Nov 12, 2018 at 5:10 AM Zong Li <zongbox@gmail.com> wrote: > > > > The stat64 family that is used on 32-bit architectures to replace > > newstat. > > > > Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request > > newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' > > > > Signed-off-by: Zong Li <zong@andestech.com> > > --- > > arch/riscv/include/asm/unistd.h | 1 + > > 1 file changed, 1 insertion(+) > > > > diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h > > index eff7aa9..a4aade9 100644 > > --- a/arch/riscv/include/asm/unistd.h > > +++ b/arch/riscv/include/asm/unistd.h > > @@ -18,5 +18,6 @@ > > > > #define __ARCH_WANT_NEW_STAT > > #define __ARCH_WANT_SYS_CLONE > > +#define __ARCH_WANT_STAT64 > > #include <uapi/asm/unistd.h> > > #include <uapi/asm/syscalls.h> > > -- > > See: http://lists.infradead.org/pipermail/linux-riscv/2018-November/002087.html > > The plan is not to have old stat syscalls and support statx on > riscv32, which is y2038 safe. > > The issue you see is a bug in include/uapi/asm-generic/unistd.h. > Marcin (CC) already sent a patch to Arnd (CC) IIRC. Basically without > __ARCH_WANT_NEW_STAT or __ARCH_WANT_STAT64 two macros are not defined: > __NR3264_fstatat and __NR3264_fstat. Which is later used (without any > guards): > > 763 #define __NR_newfstatat __NR3264_fstatat > 764 #define __NR_fstat __NR3264_fstat Nice. Thank you. ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 8:29 ` Zong Li @ 2018-11-12 8:29 ` Zong Li 0 siblings, 0 replies; 12+ messages in thread From: Zong Li @ 2018-11-12 8:29 UTC (permalink / raw) To: david.abdurachmanov Cc: marcin.juszkiewicz, aou, Arnd Bergmann, Palmer Dabbelt, Linux Kernel Mailing List, Zong Li, linux-riscv David Abdurachmanov <david.abdurachmanov@gmail.com> 於 2018年11月12日 週一 下午2:19寫道： > > On Mon, Nov 12, 2018 at 5:10 AM Zong Li <zongbox@gmail.com> wrote: > > > > The stat64 family that is used on 32-bit architectures to replace > > newstat. > > > > Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request > > newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' > > > > Signed-off-by: Zong Li <zong@andestech.com> > > --- > > arch/riscv/include/asm/unistd.h | 1 + > > 1 file changed, 1 insertion(+) > > > > diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h > > index eff7aa9..a4aade9 100644 > > --- a/arch/riscv/include/asm/unistd.h > > +++ b/arch/riscv/include/asm/unistd.h > > @@ -18,5 +18,6 @@ > > > > #define __ARCH_WANT_NEW_STAT > > #define __ARCH_WANT_SYS_CLONE > > +#define __ARCH_WANT_STAT64 > > #include <uapi/asm/unistd.h> > > #include <uapi/asm/syscalls.h> > > -- > > See: http://lists.infradead.org/pipermail/linux-riscv/2018-November/002087.html > > The plan is not to have old stat syscalls and support statx on > riscv32, which is y2038 safe. > > The issue you see is a bug in include/uapi/asm-generic/unistd.h. > Marcin (CC) already sent a patch to Arnd (CC) IIRC. Basically without > __ARCH_WANT_NEW_STAT or __ARCH_WANT_STAT64 two macros are not defined: > __NR3264_fstatat and __NR3264_fstat. Which is later used (without any > guards): > > 763 #define __NR_newfstatat __NR3264_fstatat > 764 #define __NR_fstat __NR3264_fstat Nice. Thank you. _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv ^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 8:29 ` Zong Li @ 2018-11-12 17:01 ` Palmer Dabbelt 2018-11-12 17:01 ` Palmer Dabbelt 2 siblings, 1 reply; 12+ messages in thread From: Palmer Dabbelt @ 2018-11-12 17:01 UTC (permalink / raw) To: linux-riscv On Sun, 11 Nov 2018 22:19:02 PST (-0800), david.abdurachmanov at gmail.com wrote: > On Mon, Nov 12, 2018 at 5:10 AM Zong Li <zongbox@gmail.com> wrote: >> >> The stat64 family that is used on 32-bit architectures to replace >> newstat. >> >> Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request >> newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' >> >> Signed-off-by: Zong Li <zong@andestech.com> >> --- >> arch/riscv/include/asm/unistd.h | 1 + >> 1 file changed, 1 insertion(+) >> >> diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h >> index eff7aa9..a4aade9 100644 >> --- a/arch/riscv/include/asm/unistd.h >> +++ b/arch/riscv/include/asm/unistd.h >> @@ -18,5 +18,6 @@ >> >> #define __ARCH_WANT_NEW_STAT >> #define __ARCH_WANT_SYS_CLONE >> +#define __ARCH_WANT_STAT64 >> #include <uapi/asm/unistd.h> >> #include <uapi/asm/syscalls.h> >> -- > > See: http://lists.infradead.org/pipermail/linux-riscv/2018-November/002087.html > > The plan is not to have old stat syscalls and support statx on > riscv32, which is y2038 safe. > > The issue you see is a bug in include/uapi/asm-generic/unistd.h. > Marcin (CC) already sent a patch to Arnd (CC) IIRC. Basically without > __ARCH_WANT_NEW_STAT or __ARCH_WANT_STAT64 two macros are not defined: > __NR3264_fstatat and __NR3264_fstat. Which is later used (without any > guards): > > 763 #define __NR_newfstatat __NR3264_fstatat > 764 #define __NR_fstat __NR3264_fstat I agree. Thanks! ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 1/2] RISC-V: Request stat64 on RV32 2018-11-12 17:01 ` Palmer Dabbelt @ 2018-11-12 17:01 ` Palmer Dabbelt 0 siblings, 0 replies; 12+ messages in thread From: Palmer Dabbelt @ 2018-11-12 17:01 UTC (permalink / raw) To: david.abdurachmanov Cc: marcin.juszkiewicz, aou, Arnd Bergmann, zongbox, linux-kernel, zong, linux-riscv On Sun, 11 Nov 2018 22:19:02 PST (-0800), david.abdurachmanov@gmail.com wrote: > On Mon, Nov 12, 2018 at 5:10 AM Zong Li <zongbox@gmail.com> wrote: >> >> The stat64 family that is used on 32-bit architectures to replace >> newstat. >> >> Since commit 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 ("RISC-V: Request >> newstat syscalls"), the RV32 build fail with undeclared 'sys_fstatat64' >> >> Signed-off-by: Zong Li <zong@andestech.com> >> --- >> arch/riscv/include/asm/unistd.h | 1 + >> 1 file changed, 1 insertion(+) >> >> diff --git a/arch/riscv/include/asm/unistd.h b/arch/riscv/include/asm/unistd.h >> index eff7aa9..a4aade9 100644 >> --- a/arch/riscv/include/asm/unistd.h >> +++ b/arch/riscv/include/asm/unistd.h >> @@ -18,5 +18,6 @@ >> >> #define __ARCH_WANT_NEW_STAT >> #define __ARCH_WANT_SYS_CLONE >> +#define __ARCH_WANT_STAT64 >> #include <uapi/asm/unistd.h> >> #include <uapi/asm/syscalls.h> >> -- > > See: http://lists.infradead.org/pipermail/linux-riscv/2018-November/002087.html > > The plan is not to have old stat syscalls and support statx on > riscv32, which is y2038 safe. > > The issue you see is a bug in include/uapi/asm-generic/unistd.h. > Marcin (CC) already sent a patch to Arnd (CC) IIRC. Basically without > __ARCH_WANT_NEW_STAT or __ARCH_WANT_STAT64 two macros are not defined: > __NR3264_fstatat and __NR3264_fstat. Which is later used (without any > guards): > > 763 #define __NR_newfstatat __NR3264_fstatat > 764 #define __NR_fstat __NR3264_fstat I agree. Thanks! _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv ^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH 2/2] RISC-V: Support MODULE_SECTIONS mechanism on RV32 2018-11-12 4:07 [PATCH 0/2] Fix some issue on RV32 Zong Li 2018-11-12 4:07 ` Zong Li 2018-11-12 4:07 ` [PATCH 1/2] RISC-V: Request stat64 " Zong Li @ 2018-11-12 4:07 ` Zong Li 2018-11-12 4:07 ` Zong Li 2 siblings, 1 reply; 12+ messages in thread From: Zong Li @ 2018-11-12 4:07 UTC (permalink / raw) To: linux-riscv This patch supports dynamic generate got and plt sections mechanism on rv32. It contains the modification as follows: - Always enable MODULE_SECTIONS (both rv64 and rv32) - Change the fixed size type. This patch had been tested by following modules: btrfs 6795991 0 - Live 0xa544b000 test_static_keys 17304 0 - Live 0xa28be000 zstd_compress 1198986 1 btrfs, Live 0xa2a25000 zstd_decompress 608112 1 btrfs, Live 0xa24e7000 lzo 8787 0 - Live 0xa2049000 xor 27461 1 btrfs, Live 0xa2041000 zram 78849 0 - Live 0xa2276000 netdevsim 55909 0 - Live 0xa202d000 tun 211534 0 - Live 0xa21b5000 fuse 566049 0 - Live 0xa25fb000 nfs_layout_flexfiles 192597 0 - Live 0xa229b000 ramoops 74895 0 - Live 0xa2019000 xfs 3973221 0 - Live 0xa507f000 libcrc32c 3053 2 btrfs,xfs, Live 0xa34af000 lzo_compress 17302 2 btrfs,lzo, Live 0xa347d000 lzo_decompress 7178 2 btrfs,lzo, Live 0xa3451000 raid6_pq 142086 1 btrfs, Live 0xa33a4000 reed_solomon 31022 1 ramoops, Live 0xa31eb000 test_bitmap 3734 0 - Live 0xa31af000 test_bpf 1588736 0 - Live 0xa2c11000 test_kmod 41161 0 - Live 0xa29f8000 test_module 1356 0 - Live 0xa299e000 test_printf 6024 0 [permanent], Live 0xa2971000 test_static_key_base 5797 1 test_static_keys, Live 0xa2931000 test_user_copy 4382 0 - Live 0xa28c9000 xxhash 70501 2 zstd_compress,zstd_decompress, Live 0xa2055000 Signed-off-by: Zong Li <zong@andestech.com> --- arch/riscv/Kconfig | 2 +- arch/riscv/include/asm/module.h | 28 +++++++++++++++------------- arch/riscv/kernel/module-sections.c | 30 ++++++++++++++++-------------- 3 files changed, 32 insertions(+), 28 deletions(-) diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig index 55da93f..477f1e8 100644 --- a/arch/riscv/Kconfig +++ b/arch/riscv/Kconfig @@ -39,6 +39,7 @@ config RISCV select SYSCTL_EXCEPTION_TRACE select HAVE_ARCH_TRACEHOOK select MODULES_USE_ELF_RELA if MODULES + select MODULE_SECTIONS if MODULES select THREAD_INFO_IN_TASK select RISCV_TIMER select GENERIC_IRQ_MULTI_HANDLER @@ -149,7 +150,6 @@ choice bool "2GiB" config MAXPHYSMEM_128GB depends on 64BIT && CMODEL_MEDANY - select MODULE_SECTIONS if MODULES bool "128GiB" endchoice diff --git a/arch/riscv/include/asm/module.h b/arch/riscv/include/asm/module.h index 349df33..27328fe 100644 --- a/arch/riscv/include/asm/module.h +++ b/arch/riscv/include/asm/module.h @@ -8,12 +8,12 @@ #define MODULE_ARCH_VERMAGIC "riscv" -u64 module_emit_got_entry(struct module *mod, u64 val); -u64 module_emit_plt_entry(struct module *mod, u64 val); +unsigned long module_emit_got_entry(struct module *mod, unsigned long val); +unsigned long module_emit_plt_entry(struct module *mod, unsigned long val); #ifdef CONFIG_MODULE_SECTIONS struct mod_section { - struct elf64_shdr *shdr; + Elf_Shdr *shdr; int num_entries; int max_entries; }; @@ -25,18 +25,18 @@ struct mod_arch_specific { }; struct got_entry { - u64 symbol_addr; /* the real variable address */ + unsigned long symbol_addr; /* the real variable address */ }; -static inline struct got_entry emit_got_entry(u64 val) +static inline struct got_entry emit_got_entry(unsigned long val) { return (struct got_entry) {val}; } -static inline struct got_entry *get_got_entry(u64 val, +static inline struct got_entry *get_got_entry(unsigned long val, const struct mod_section *sec) { - struct got_entry *got = (struct got_entry *)sec->shdr->sh_addr; + struct got_entry *got = (struct got_entry *)(sec->shdr->sh_addr); int i; for (i = 0; i < sec->num_entries; i++) { if (got[i].symbol_addr == val) @@ -61,7 +61,9 @@ struct plt_entry { #define REG_T0 0x5 #define REG_T1 0x6 -static inline struct plt_entry emit_plt_entry(u64 val, u64 plt, u64 got_plt) +static inline struct plt_entry emit_plt_entry(unsigned long val, + unsigned long plt, + unsigned long got_plt) { /* * U-Type encoding: @@ -75,7 +77,7 @@ static inline struct plt_entry emit_plt_entry(u64 val, u64 plt, u64 got_plt) * +------------+------------+--------+----------+----------+ * */ - u64 offset = got_plt - plt; + unsigned long offset = got_plt - plt; u32 hi20 = (offset + 0x800) & 0xfffff000; u32 lo12 = (offset - hi20); return (struct plt_entry) { @@ -85,7 +87,7 @@ static inline struct plt_entry emit_plt_entry(u64 val, u64 plt, u64 got_plt) }; } -static inline int get_got_plt_idx(u64 val, const struct mod_section *sec) +static inline int get_got_plt_idx(unsigned long val, const struct mod_section *sec) { struct got_entry *got_plt = (struct got_entry *)sec->shdr->sh_addr; int i; @@ -96,9 +98,9 @@ static inline int get_got_plt_idx(u64 val, const struct mod_section *sec) return -1; } -static inline struct plt_entry *get_plt_entry(u64 val, - const struct mod_section *sec_plt, - const struct mod_section *sec_got_plt) +static inline struct plt_entry *get_plt_entry(unsigned long val, + const struct mod_section *sec_plt, + const struct mod_section *sec_got_plt) { struct plt_entry *plt = (struct plt_entry *)sec_plt->shdr->sh_addr; int got_plt_idx = get_got_plt_idx(val, sec_got_plt); diff --git a/arch/riscv/kernel/module-sections.c b/arch/riscv/kernel/module-sections.c index bbbd26e..e8b991a 100644 --- a/arch/riscv/kernel/module-sections.c +++ b/arch/riscv/kernel/module-sections.c @@ -9,14 +9,14 @@ #include <linux/kernel.h> #include <linux/module.h> -u64 module_emit_got_entry(struct module *mod, u64 val) +unsigned long module_emit_got_entry(struct module *mod, unsigned long val) { struct mod_section *got_sec = &mod->arch.got; int i = got_sec->num_entries; struct got_entry *got = get_got_entry(val, got_sec); if (got) - return (u64)got; + return (unsigned long)got; /* There is no duplicate entry, create a new one */ got = (struct got_entry *)got_sec->shdr->sh_addr; @@ -25,10 +25,10 @@ u64 module_emit_got_entry(struct module *mod, u64 val) got_sec->num_entries++; BUG_ON(got_sec->num_entries > got_sec->max_entries); - return (u64)&got[i]; + return (unsigned long)&got[i]; } -u64 module_emit_plt_entry(struct module *mod, u64 val) +unsigned long module_emit_plt_entry(struct module *mod, unsigned long val) { struct mod_section *got_plt_sec = &mod->arch.got_plt; struct got_entry *got_plt; @@ -37,27 +37,29 @@ u64 module_emit_plt_entry(struct module *mod, u64 val) int i = plt_sec->num_entries; if (plt) - return (u64)plt; + return (unsigned long)plt; /* There is no duplicate entry, create a new one */ got_plt = (struct got_entry *)got_plt_sec->shdr->sh_addr; got_plt[i] = emit_got_entry(val); plt = (struct plt_entry *)plt_sec->shdr->sh_addr; - plt[i] = emit_plt_entry(val, (u64)&plt[i], (u64)&got_plt[i]); + plt[i] = emit_plt_entry(val, + (unsigned long)&plt[i], + (unsigned long)&got_plt[i]); plt_sec->num_entries++; got_plt_sec->num_entries++; BUG_ON(plt_sec->num_entries > plt_sec->max_entries); - return (u64)&plt[i]; + return (unsigned long)&plt[i]; } -static int is_rela_equal(const Elf64_Rela *x, const Elf64_Rela *y) +static int is_rela_equal(const Elf_Rela *x, const Elf_Rela *y) { return x->r_info == y->r_info && x->r_addend == y->r_addend; } -static bool duplicate_rela(const Elf64_Rela *rela, int idx) +static bool duplicate_rela(const Elf_Rela *rela, int idx) { int i; for (i = 0; i < idx; i++) { @@ -67,13 +69,13 @@ static bool duplicate_rela(const Elf64_Rela *rela, int idx) return false; } -static void count_max_entries(Elf64_Rela *relas, int num, +static void count_max_entries(Elf_Rela *relas, int num, unsigned int *plts, unsigned int *gots) { unsigned int type, i; for (i = 0; i < num; i++) { - type = ELF64_R_TYPE(relas[i].r_info); + type = ELF_RISCV_R_TYPE(relas[i].r_info); if (type == R_RISCV_CALL_PLT) { if (!duplicate_rela(relas, i)) (*plts)++; @@ -118,9 +120,9 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs, /* Calculate the maxinum number of entries */ for (i = 0; i < ehdr->e_shnum; i++) { - Elf64_Rela *relas = (void *)ehdr + sechdrs[i].sh_offset; - int num_rela = sechdrs[i].sh_size / sizeof(Elf64_Rela); - Elf64_Shdr *dst_sec = sechdrs + sechdrs[i].sh_info; + Elf_Rela *relas = (void *)ehdr + sechdrs[i].sh_offset; + int num_rela = sechdrs[i].sh_size / sizeof(Elf_Rela); + Elf_Shdr *dst_sec = sechdrs + sechdrs[i].sh_info; if (sechdrs[i].sh_type != SHT_RELA) continue; -- 2.7.4 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [PATCH 2/2] RISC-V: Support MODULE_SECTIONS mechanism on RV32 2018-11-12 4:07 ` [PATCH 2/2] RISC-V: Support MODULE_SECTIONS mechanism " Zong Li @ 2018-11-12 4:07 ` Zong Li 0 siblings, 0 replies; 12+ messages in thread From: Zong Li @ 2018-11-12 4:07 UTC (permalink / raw) To: palmer, aou, linux-riscv, linux-kernel, zong; +Cc: Zong Li This patch supports dynamic generate got and plt sections mechanism on rv32. It contains the modification as follows: - Always enable MODULE_SECTIONS (both rv64 and rv32) - Change the fixed size type. This patch had been tested by following modules: btrfs 6795991 0 - Live 0xa544b000 test_static_keys 17304 0 - Live 0xa28be000 zstd_compress 1198986 1 btrfs, Live 0xa2a25000 zstd_decompress 608112 1 btrfs, Live 0xa24e7000 lzo 8787 0 - Live 0xa2049000 xor 27461 1 btrfs, Live 0xa2041000 zram 78849 0 - Live 0xa2276000 netdevsim 55909 0 - Live 0xa202d000 tun 211534 0 - Live 0xa21b5000 fuse 566049 0 - Live 0xa25fb000 nfs_layout_flexfiles 192597 0 - Live 0xa229b000 ramoops 74895 0 - Live 0xa2019000 xfs 3973221 0 - Live 0xa507f000 libcrc32c 3053 2 btrfs,xfs, Live 0xa34af000 lzo_compress 17302 2 btrfs,lzo, Live 0xa347d000 lzo_decompress 7178 2 btrfs,lzo, Live 0xa3451000 raid6_pq 142086 1 btrfs, Live 0xa33a4000 reed_solomon 31022 1 ramoops, Live 0xa31eb000 test_bitmap 3734 0 - Live 0xa31af000 test_bpf 1588736 0 - Live 0xa2c11000 test_kmod 41161 0 - Live 0xa29f8000 test_module 1356 0 - Live 0xa299e000 test_printf 6024 0 [permanent], Live 0xa2971000 test_static_key_base 5797 1 test_static_keys, Live 0xa2931000 test_user_copy 4382 0 - Live 0xa28c9000 xxhash 70501 2 zstd_compress,zstd_decompress, Live 0xa2055000 Signed-off-by: Zong Li <zong@andestech.com> --- arch/riscv/Kconfig | 2 +- arch/riscv/include/asm/module.h | 28 +++++++++++++++------------- arch/riscv/kernel/module-sections.c | 30 ++++++++++++++++-------------- 3 files changed, 32 insertions(+), 28 deletions(-) diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig index 55da93f..477f1e8 100644 --- a/arch/riscv/Kconfig +++ b/arch/riscv/Kconfig @@ -39,6 +39,7 @@ config RISCV select SYSCTL_EXCEPTION_TRACE select HAVE_ARCH_TRACEHOOK select MODULES_USE_ELF_RELA if MODULES + select MODULE_SECTIONS if MODULES select THREAD_INFO_IN_TASK select RISCV_TIMER select GENERIC_IRQ_MULTI_HANDLER @@ -149,7 +150,6 @@ choice bool "2GiB" config MAXPHYSMEM_128GB depends on 64BIT && CMODEL_MEDANY - select MODULE_SECTIONS if MODULES bool "128GiB" endchoice diff --git a/arch/riscv/include/asm/module.h b/arch/riscv/include/asm/module.h index 349df33..27328fe 100644 --- a/arch/riscv/include/asm/module.h +++ b/arch/riscv/include/asm/module.h @@ -8,12 +8,12 @@ #define MODULE_ARCH_VERMAGIC "riscv" -u64 module_emit_got_entry(struct module *mod, u64 val); -u64 module_emit_plt_entry(struct module *mod, u64 val); +unsigned long module_emit_got_entry(struct module *mod, unsigned long val); +unsigned long module_emit_plt_entry(struct module *mod, unsigned long val); #ifdef CONFIG_MODULE_SECTIONS struct mod_section { - struct elf64_shdr *shdr; + Elf_Shdr *shdr; int num_entries; int max_entries; }; @@ -25,18 +25,18 @@ struct mod_arch_specific { }; struct got_entry { - u64 symbol_addr; /* the real variable address */ + unsigned long symbol_addr; /* the real variable address */ }; -static inline struct got_entry emit_got_entry(u64 val) +static inline struct got_entry emit_got_entry(unsigned long val) { return (struct got_entry) {val}; } -static inline struct got_entry *get_got_entry(u64 val, +static inline struct got_entry *get_got_entry(unsigned long val, const struct mod_section *sec) { - struct got_entry *got = (struct got_entry *)sec->shdr->sh_addr; + struct got_entry *got = (struct got_entry *)(sec->shdr->sh_addr); int i; for (i = 0; i < sec->num_entries; i++) { if (got[i].symbol_addr == val) @@ -61,7 +61,9 @@ struct plt_entry { #define REG_T0 0x5 #define REG_T1 0x6 -static inline struct plt_entry emit_plt_entry(u64 val, u64 plt, u64 got_plt) +static inline struct plt_entry emit_plt_entry(unsigned long val, + unsigned long plt, + unsigned long got_plt) { /* * U-Type encoding: @@ -75,7 +77,7 @@ static inline struct plt_entry emit_plt_entry(u64 val, u64 plt, u64 got_plt) * +------------+------------+--------+----------+----------+ * */ - u64 offset = got_plt - plt; + unsigned long offset = got_plt - plt; u32 hi20 = (offset + 0x800) & 0xfffff000; u32 lo12 = (offset - hi20); return (struct plt_entry) { @@ -85,7 +87,7 @@ static inline struct plt_entry emit_plt_entry(u64 val, u64 plt, u64 got_plt) }; } -static inline int get_got_plt_idx(u64 val, const struct mod_section *sec) +static inline int get_got_plt_idx(unsigned long val, const struct mod_section *sec) { struct got_entry *got_plt = (struct got_entry *)sec->shdr->sh_addr; int i; @@ -96,9 +98,9 @@ static inline int get_got_plt_idx(u64 val, const struct mod_section *sec) return -1; } -static inline struct plt_entry *get_plt_entry(u64 val, - const struct mod_section *sec_plt, - const struct mod_section *sec_got_plt) +static inline struct plt_entry *get_plt_entry(unsigned long val, + const struct mod_section *sec_plt, + const struct mod_section *sec_got_plt) { struct plt_entry *plt = (struct plt_entry *)sec_plt->shdr->sh_addr; int got_plt_idx = get_got_plt_idx(val, sec_got_plt); diff --git a/arch/riscv/kernel/module-sections.c b/arch/riscv/kernel/module-sections.c index bbbd26e..e8b991a 100644 --- a/arch/riscv/kernel/module-sections.c +++ b/arch/riscv/kernel/module-sections.c @@ -9,14 +9,14 @@ #include <linux/kernel.h> #include <linux/module.h> -u64 module_emit_got_entry(struct module *mod, u64 val) +unsigned long module_emit_got_entry(struct module *mod, unsigned long val) { struct mod_section *got_sec = &mod->arch.got; int i = got_sec->num_entries; struct got_entry *got = get_got_entry(val, got_sec); if (got) - return (u64)got; + return (unsigned long)got; /* There is no duplicate entry, create a new one */ got = (struct got_entry *)got_sec->shdr->sh_addr; @@ -25,10 +25,10 @@ u64 module_emit_got_entry(struct module *mod, u64 val) got_sec->num_entries++; BUG_ON(got_sec->num_entries > got_sec->max_entries); - return (u64)&got[i]; + return (unsigned long)&got[i]; } -u64 module_emit_plt_entry(struct module *mod, u64 val) +unsigned long module_emit_plt_entry(struct module *mod, unsigned long val) { struct mod_section *got_plt_sec = &mod->arch.got_plt; struct got_entry *got_plt; @@ -37,27 +37,29 @@ u64 module_emit_plt_entry(struct module *mod, u64 val) int i = plt_sec->num_entries; if (plt) - return (u64)plt; + return (unsigned long)plt; /* There is no duplicate entry, create a new one */ got_plt = (struct got_entry *)got_plt_sec->shdr->sh_addr; got_plt[i] = emit_got_entry(val); plt = (struct plt_entry *)plt_sec->shdr->sh_addr; - plt[i] = emit_plt_entry(val, (u64)&plt[i], (u64)&got_plt[i]); + plt[i] = emit_plt_entry(val, + (unsigned long)&plt[i], + (unsigned long)&got_plt[i]); plt_sec->num_entries++; got_plt_sec->num_entries++; BUG_ON(plt_sec->num_entries > plt_sec->max_entries); - return (u64)&plt[i]; + return (unsigned long)&plt[i]; } -static int is_rela_equal(const Elf64_Rela *x, const Elf64_Rela *y) +static int is_rela_equal(const Elf_Rela *x, const Elf_Rela *y) { return x->r_info == y->r_info && x->r_addend == y->r_addend; } -static bool duplicate_rela(const Elf64_Rela *rela, int idx) +static bool duplicate_rela(const Elf_Rela *rela, int idx) { int i; for (i = 0; i < idx; i++) { @@ -67,13 +69,13 @@ static bool duplicate_rela(const Elf64_Rela *rela, int idx) return false; } -static void count_max_entries(Elf64_Rela *relas, int num, +static void count_max_entries(Elf_Rela *relas, int num, unsigned int *plts, unsigned int *gots) { unsigned int type, i; for (i = 0; i < num; i++) { - type = ELF64_R_TYPE(relas[i].r_info); + type = ELF_RISCV_R_TYPE(relas[i].r_info); if (type == R_RISCV_CALL_PLT) { if (!duplicate_rela(relas, i)) (*plts)++; @@ -118,9 +120,9 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs, /* Calculate the maxinum number of entries */ for (i = 0; i < ehdr->e_shnum; i++) { - Elf64_Rela *relas = (void *)ehdr + sechdrs[i].sh_offset; - int num_rela = sechdrs[i].sh_size / sizeof(Elf64_Rela); - Elf64_Shdr *dst_sec = sechdrs + sechdrs[i].sh_info; + Elf_Rela *relas = (void *)ehdr + sechdrs[i].sh_offset; + int num_rela = sechdrs[i].sh_size / sizeof(Elf_Rela); + Elf_Shdr *dst_sec = sechdrs + sechdrs[i].sh_info; if (sechdrs[i].sh_type != SHT_RELA) continue; -- 2.7.4 _______________________________________________ linux-riscv mailing list linux-riscv@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv ^ permalink raw reply related [flat|nested] 12+ messages in thread
end of thread, other threads:[~2018-11-12 17:02 UTC | newest] Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2018-11-12 4:07 [PATCH 0/2] Fix some issue on RV32 Zong Li 2018-11-12 4:07 ` Zong Li 2018-11-12 4:07 ` [PATCH 1/2] RISC-V: Request stat64 " Zong Li 2018-11-12 4:07 ` Zong Li 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 6:19 ` David Abdurachmanov 2018-11-12 8:29 ` Zong Li 2018-11-12 8:29 ` Zong Li 2018-11-12 17:01 ` Palmer Dabbelt 2018-11-12 17:01 ` Palmer Dabbelt 2018-11-12 4:07 ` [PATCH 2/2] RISC-V: Support MODULE_SECTIONS mechanism " Zong Li 2018-11-12 4:07 ` Zong Li
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).