riscv: improve __ex_table section handling
cover riscv: improve __ex_table section handling
From: Jisheng Zhang <[email protected]>
Enable BUILDTIME_TABLE_SORT to sort the exception table at build time then move exception table to RO_DATA segment.
Jisheng Zhang (2):
riscv: Enable BUILDTIME_TABLE_SORT
riscv: Move EXCEPTION_TABLE to RO_DATA segment
arch/riscv/Kconfig | 1 + arch/riscv/kernel/vmlinux-xip.lds.S | 1 - arch/riscv/kernel/vmlinux.lds.S | 4 ++-- scripts/sorttable.c | 1 + 4 files changed, 4 insertions(+), 3 deletions(-)
Signature · 8 lines
2.32.0 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
PATCH 1/2 riscv: Enable BUILDTIME_TABLE_SORT
From: Jisheng Zhang <[email protected]>
Enable BUILDTIME_TABLE_SORT to sort the exception table at build time rather than during boot.
--- arch/riscv/Kconfig | 1 + scripts/sorttable.c | 1 + 2 files changed, 2 insertions(+)
arch/riscv/Kconfig +1 -0
@@ -40,6 +40,7 @@ config RISCV select ARCH_WANT_FRAME_POINTERS select ARCH_WANT_HUGE_PMD_SHARE if 64BIT select BINFMT_FLAT_NO_DATA_START_OFFSET if !MMU + select BUILDTIME_TABLE_SORT select CLONE_BACKWARDS select CLINT_TIMER if !MMU select COMMON_CLK
scripts/sorttable.c +1 -0
@@ -349,6 +349,7 @@ static int do_file(char const *const fname, void *addr)case EM_ARM :case EM_MICROBLAZE :case EM_MIPS : +case EM_RISCV :case EM_XTENSA :break ;default :
Signature · 8 lines
2.32.0 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
PATCH 2/2 riscv: Move EXCEPTION_TABLE to RO_DATA segment
In reply to Jisheng ZhangFrom: Jisheng Zhang <[email protected]>
The _ex_table section is read-only, so move it to RO_DATA.
--- arch/riscv/kernel/vmlinux-xip.lds.S | 1 - arch/riscv/kernel/vmlinux.lds.S | 4 ++-- 2 files changed, 2 insertions(+), 3 deletions(-)
arch/riscv/kernel/vmlinux-xip.lds.S +0 -1
@@ -121,7 +121,6 @@ SECTIONS } BSS_SECTION(PAGE_SIZE, PAGE_SIZE, 0) - EXCEPTION_TABLE(0x10) .rel.dyn : AT(ADDR(.rel.dyn) - LOAD_OFFSET) { *(.rel.dyn*)
arch/riscv/kernel/vmlinux.lds.S +2 -2
@@ -4,6 +4,8 @@ * Copyright (C) 2017 SiFive */ +#define RO_EXCEPTION_TABLE_ALIGN 16 + #ifdef CONFIG_XIP_KERNEL #include "vmlinux-xip.lds.S" #else
@@ -112,8 +114,6 @@ SECTIONS *(.srodata*) } - EXCEPTION_TABLE(0x10) - . = ALIGN(SECTION_ALIGN); _data = .;
Signature · 8 lines
2.32.0 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
Palmer Dabbelt
maintainerQuoted 9 lines · On Sun, 08 Aug 2021 10:25:09 PDT (-0700), [email protected] wrote:
From: Jisheng Zhang <[email protected]>
Enable
BUILDTIME_TABLE_SORTto sort the exception table at build time then move exception table toRO_DATAsegment.Jisheng Zhang (2):
riscv: EnableBUILDTIME_TABLE_SORT
riscv: MoveEXCEPTION_TABLEtoRO_DATAsegment
Quoted 5 lines
arch/riscv/Kconfig | 1 + arch/riscv/kernel/vmlinux-xip.lds.S | 1 - arch/riscv/kernel/vmlinux.lds.S | 4 ++-- scripts/sorttable.c | 1 + 4 files changed, 4 insertions(+), 3 deletions(-)
This seems reasonable, but it's failing for some configurations (at least tinyconfig) saying there is no __ex_table. I'm not entirely sure how that comes about, as we've got them for futexes and uaccess.
Maybe the right thing to do here is to fix scripts/sorttable.c so it can handle files with nothing to sort? I think it's just as simple as a successful early out like this
scripts/sorttable.h +0 -0
@@ -294,8 +294,9 @@ static int do_sort(Elf_Ehdr *ehdr, goto out; }
#endif
+ /* If there is no __ex_table section there is no work do to. */
if (!extab_sec) {
- fprintf(stderr, "no __ex_table in file: %s\n", fname);
+ rc = 0;
goto out;
}I'm not entirely sure though -- my logic is essentially just "there's no __ex_table, so there's nothing to sort, so just don't try".
All the configurations I can actually boot have an __ex_table, so I'm not sure how to test that.
Quoted 10 lines · On Wed, 25 Aug 2021 23:47:02 -0700 (PDT) Palmer Dabbelt <[email protected]>…
Quoted 9 lines · On Sun, 08 Aug 2021 10:25:09 PDT (-0700), [email protected] wrote:
From: Jisheng Zhang <[email protected]>
Enable
BUILDTIME_TABLE_SORTto sort the exception table at build time then move exception table toRO_DATAsegment.Jisheng Zhang (2):
riscv: EnableBUILDTIME_TABLE_SORT
riscv: MoveEXCEPTION_TABLEtoRO_DATAsegment
Quoted 8 lines
Quoted 5 lines
arch/riscv/Kconfig | 1 + arch/riscv/kernel/vmlinux-xip.lds.S | 1 - arch/riscv/kernel/vmlinux.lds.S | 4 ++-- scripts/sorttable.c | 1 + 4 files changed, 4 insertions(+), 3 deletions(-)This seems reasonable, but it's failing for some configurations (at least tinyconfig) saying there is no
__ex_table. I'm not entirely sure
Nice catch! tinyconfig in riscv means NOMMU which indicates no __ex_table at all. It seems we have to only enable BUILDTIME_TABLE_SORT for MMU case.
will send out V2 soon.
Thanks
Quoted 24 lines
how that comes about, as we've got them for futexes and uaccess.
Maybe the right thing to do here is to fix
scripts/sorttable.cso it can handle files with nothing to sort? I think it's just as simple as a successful early out like thisscripts/sorttable.h +2
-1@@ -294,8 +294,9 @@ static int do_sort(Elf_Ehdr *ehdr, goto out; }#endif +/* If there is no __ex_table section there is no work do to. */ if (!extab_sec) { -fprintf (stderr,"no __ex_table in file: %s\n" , fname); + rc =0 ; goto out; }I'm not entirely sure though -- my logic is essentially just "there's no
__ex_table, so there's nothing to sort, so just don't try".
Quoted 7 lines
All the configurations I can actually boot have an
__ex_table, so I'm not sure how to test that._______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv