| Message ID | 1562843158-31390-1-git-send-email-mihailo.stojanovic@rt-rk.com |
|---|---|
| State | New |
| Headers | show |
| Series | [v2] MIPS support for GNU hash | expand |
Hi, I would like to check the status of this patch, given the fact that it's a release blocker for the glibc 2.30. Cheers, Mihailo
On 11/07/2019 08:05, Mihailo Stojanovic wrote: > Hello everyone, > > This patch is a reimplementation of [1], which was submitted back in > 2015. Copyright issue has been sorted out [2] last year. It proposed a > new section (.gnu.xhash) and related dynamic tag (DT_GNU_XHASH). The new > section would be virtually identical to the existing .gnu.hash except > for the translation table (xlat) which would contain correct MIPS > .dynsym indexes corresponding to the hashvals in chains. This is because > MIPS ABI imposes a different ordering of the dynyms than the one > expected by the .gnu.hash section. Another addition would be a leading > word which would contain the number of entries in the translation table. > > In this patch, the new section name and dynamic tag are changed to > reflect the fact that the section should be treated as MIPS-specific > (.MIPS.xhash and DT_MIPS_XHASH). > > This patch addresses the alignment issue as reported in [3] which is > caused by the leading word added to the .MIPS.xhash section. Leading word > is now removed in the corresponding binutils patch, and the number of > entries in the translation table is computed using DT_MIPS_SYMTABNO > dynamic tag. > > Since the MIPS specific dl-lookup.c file was removed since the initial > patch, I opted for the definition of three new macros in the generic > ldsodefs.h. ELF_MACHINE_GNU_HASH_ADDRIDX defines the index of the > dynamic tag in the l_info array. ELF_MACHINE_HASH_SYMIDX is used to > calculate the index of a symbol in GNU hash. On MIPS, it is defined to > look up the symbol index in the translation table. > ELF_MACHINE_XHASH_SETUP is defined for MIPS only. It initializes the > .MIPS.xhash pointer in the link_map_machine struct. > > The other major change is reserving MIPS ABI version 5 for .MIPS.xhash, > suggesting that the dynamic linker now supports .MIPS.xhash. This is > something of which I am not sure, especially after reading [4]. I am > confused on whether this ABI version is reserved for IFUNC, or it can be > used for this purpose. > > The patch was tested by running the glibc testsuite for the three MIPS > ABIs (o32, n32 and n64). I focused the review on the changes that touch generic implementation, I did not focused on MIPS specific bits neither if the changes matches what binutils patch is aiming to do. The patch looks ok with on the change of ELF_MACHINE_XHASH_SETUP definition below. > > The corresponding binutils patch can be found here: > https://sourceware.org/ml/binutils/2019-07/msg00098.html If I understood correctly this patch has not been pushed upstream, neither in fact actually reviewed. My understanding is we require upstream support for either kernel or toolchain features before including it on glibc, although not sure about this specific feature where it really tied with binutils support. Carlos, do you think we should wait for binutils? > > Regards, > Mihailo > > [1] https://sourceware.org/ml/binutils/2015-10/msg00057.html > [2] https://sourceware.org/ml/binutils/2018-03/msg00025.html > [3] https://sourceware.org/ml/binutils/2016-01/msg00006.html > [4] https://sourceware.org/ml/libc-alpha/2016-12/msg00853.html > > * elf/dl-addr.c (determine_info): Calculate the symbol index > using the newly defined ELF_MACHINE_HASH_SYMIDX macro. > * elf/dl-lookup.c (do_lookup_x): Calculate the symbol index > using the newly defined ELF_MACHINE_HASH_SYMIDX macro. > (_dl_setup_hash): Initialize MIPS xhash translation table. > * elf/elf.h (SHT_MIPS_XHASH): New define. > (DT_MIPS_XHASH): New define. > * sysdeps/generic/ldsodefs.h (ELF_MACHINE_GNU_HASH_ADDRIDX): New > define. > (ELF_MACHINE_HASH_SYMIDX): Ditto. > *sysdeps/mips/ldsodefs.h (ELF_MACHINE_GNU_HASH_ADDRIDX): New > define. > (ELF_MACHINE_HASH_SYMIDX): Ditto. > (ELF_MACHINE_XHASH_SETUP): Ditto. > * sysdeps/mips/linkmap.h (struct link_map_machine): New member. > * sysdeps/unix/sysv/linux/mips/ldsodefs.h: Increment valid ABI > version. > * sysdeps/unix/sysv/linux/mips/libc-abis: New ABI version. > --- > elf/dl-addr.c | 6 ++---- > elf/dl-lookup.c | 12 +++++++++--- > elf/elf.h | 5 ++++- > sysdeps/generic/ldsodefs.h | 11 +++++++++++ > sysdeps/mips/ldsodefs.h | 11 +++++++++++ > sysdeps/mips/linkmap.h | 1 + > sysdeps/unix/sysv/linux/mips/ldsodefs.h | 2 +- > sysdeps/unix/sysv/linux/mips/libc-abis | 3 +++ > 8 files changed, 42 insertions(+), 9 deletions(-) > > diff --git a/elf/dl-addr.c b/elf/dl-addr.c > index 9d285d7..f504e03 100644 > --- a/elf/dl-addr.c > +++ b/elf/dl-addr.c > @@ -20,7 +20,6 @@ > #include <stddef.h> > #include <ldsodefs.h> > > - Gratuitous line deletion, I personally try to avoid such changes. > static inline void > __attribute ((always_inline)) > determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, > @@ -42,7 +41,7 @@ determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, > ElfW(Word) strtabsize = match->l_info[DT_STRSZ]->d_un.d_val; > > const ElfW(Sym) *matchsym = NULL; > - if (match->l_info[ADDRIDX (DT_GNU_HASH)] != NULL) > + if (match->l_info[ELF_MACHINE_GNU_HASH_ADDRIDX] != NULL) > { > /* We look at all symbol table entries referenced by the hash > table. */ > @@ -57,6 +56,7 @@ determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, > { > /* The hash table never references local symbols so > we can omit that test here. */ > + symndx = ELF_MACHINE_HASH_SYMIDX(match, hasharr); > if ((symtab[symndx].st_shndx != SHN_UNDEF > || symtab[symndx].st_value != 0) > && symtab[symndx].st_shndx != SHN_ABS Ok, assignment will be expanded to: symndx = hasharr - hasharr->l_gnu_chain_zero; > @@ -65,8 +65,6 @@ determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, > matchsym, addr) > && symtab[symndx].st_name < strtabsize) > matchsym = (ElfW(Sym) *) &symtab[symndx]; > - > - ++symndx; > } > while ((*hasharr++ & 1u) == 0); > } And then 'symndx' will be updated by 'hashrray' increment. > diff --git a/elf/dl-lookup.c b/elf/dl-lookup.c > index e3f42a1..d717ed3 100644 > --- a/elf/dl-lookup.c > +++ b/elf/dl-lookup.c > @@ -403,7 +403,7 @@ do_lookup_x (const char *undef_name, uint_fast32_t new_hash, > do > if (((*hasharr ^ new_hash) >> 1) == 0) > { > - symidx = hasharr - map->l_gnu_chain_zero; > + symidx = ELF_MACHINE_HASH_SYMIDX (map, hasharr); > sym = check_match (undef_name, ref, version, flags, > type_class, &symtab[symidx], symidx, > strtab, map, &versioned_sym, Ok. > @@ -937,10 +937,10 @@ _dl_setup_hash (struct link_map *map) > { > Elf_Symndx *hash; > > - if (__glibc_likely (map->l_info[ADDRIDX (DT_GNU_HASH)] != NULL)) > + if (__glibc_likely (map->l_info[ELF_MACHINE_GNU_HASH_ADDRIDX] != NULL)) > { > Elf32_Word *hash32 > - = (void *) D_PTR (map, l_info[ADDRIDX (DT_GNU_HASH)]); > + = (void *) D_PTR (map, l_info[ELF_MACHINE_GNU_HASH_ADDRIDX]); > map->l_nbuckets = *hash32++; > Elf32_Word symbias = *hash32++; > Elf32_Word bitmask_nwords = *hash32++; Ok. > @@ -955,6 +955,12 @@ _dl_setup_hash (struct link_map *map) > map->l_gnu_buckets = hash32; > hash32 += map->l_nbuckets; > map->l_gnu_chain_zero = hash32 - symbias; > + > + /* Initialize MIPS xhash translation table. */ > +#ifdef ELF_MACHINE_XHASH_SETUP > + ELF_MACHINE_XHASH_SETUP (hash32, symbias, map) > +#endif > + > return; > } > I think it is better we move away from ifdef such kind of logic where possible and just define ELF_MACHINE_XHASH_SETUP in the generic as an empty macro. > diff --git a/elf/elf.h b/elf/elf.h > index bb13c87..b53d815 100644 > --- a/elf/elf.h > +++ b/elf/elf.h > @@ -1715,6 +1715,7 @@ typedef struct > #define SHT_MIPS_EH_REGION 0x70000027 > #define SHT_MIPS_XLATE_OLD 0x70000028 > #define SHT_MIPS_PDR_EXCEPTION 0x70000029 > +#define SHT_MIPS_XHASH 0x7000002b > > /* Legal values for sh_flags field of Elf32_Shdr. */ > Ok, it matches the binutils patch. > @@ -1962,7 +1963,9 @@ typedef struct > in a PIE as it stores a relative offset from the address of the tag > rather than an absolute address. */ > #define DT_MIPS_RLD_MAP_REL 0x70000035 > -#define DT_MIPS_NUM 0x36 > +/* GNU-style hash table with xlat. */ > +#define DT_MIPS_XHASH 0x70000036 > +#define DT_MIPS_NUM 0x37 > > /* Legal values for DT_MIPS_FLAGS Elf32_Dyn entry. */ > Ditto. > diff --git a/sysdeps/generic/ldsodefs.h b/sysdeps/generic/ldsodefs.h > index b1fc5c3..e95d8cd 100644 > --- a/sysdeps/generic/ldsodefs.h > +++ b/sysdeps/generic/ldsodefs.h > @@ -47,6 +47,17 @@ __BEGIN_DECLS > #define ADDRIDX(tag) (DT_NUM + DT_THISPROCNUM + DT_VERSIONTAGNUM \ > + DT_EXTRANUM + DT_VALNUM + DT_ADDRTAGIDX (tag)) > > +/* Type of GNU hash which the machine uses. */ > +#ifndef ELF_MACHINE_GNU_HASH_ADDRIDX > +# define ELF_MACHINE_GNU_HASH_ADDRIDX ADDRIDX (DT_GNU_HASH) > +#endif > + > +/* Calculate the index of a symbol in GNU_HASH. */ > +#ifndef ELF_MACHINE_HASH_SYMIDX > +# define ELF_MACHINE_HASH_SYMIDX(map, hasharr) \ > + ((hasharr) - (map)->l_gnu_chain_zero) > +#endif > + > /* We use this macro to refer to ELF types independent of the native wordsize. > `ElfW(TYPE)' is used in place of `Elf32_TYPE' or `Elf64_TYPE'. */ > #define ELFW(type) _ElfW (ELF, __ELF_NATIVE_CLASS, type) Ok, the define if not since the correct option since ldsodefs is included in arch-specific versions with include_next. > diff --git a/sysdeps/mips/ldsodefs.h b/sysdeps/mips/ldsodefs.h > index f0acb02..0c6f027 100644 > --- a/sysdeps/mips/ldsodefs.h > +++ b/sysdeps/mips/ldsodefs.h > @@ -26,6 +26,17 @@ struct La_mips_32_retval; > struct La_mips_64_regs; > struct La_mips_64_retval; > > +#define ELF_MACHINE_GNU_HASH_ADDRIDX (DT_MIPS_XHASH - DT_LOPROC + DT_NUM) > + > +/* Calculate the index of a symbol in MIPS_XHASH. */ > +#define ELF_MACHINE_HASH_SYMIDX(map, hasharr) \ > + ((map)->l_mach.mips_xlat_zero[(hasharr) - (map)->l_gnu_chain_zero]) > + > +/* Setup MIPS_XHASH. */ > +#define ELF_MACHINE_XHASH_SETUP(hash32, symbias, map) \ > + (hash32) += (map)->l_info[DT_MIPS (SYMTABNO)]->d_un.d_val - (symbias); \ > + (map)->l_mach.mips_xlat_zero = (hash32) - (symbias); > + > #define ARCH_PLTENTER_MEMBERS \ > Elf32_Addr (*mips_o32_gnu_pltenter) (Elf32_Sym *, unsigned int, \ > uintptr_t *, uintptr_t *, \ Ok. > diff --git a/sysdeps/mips/linkmap.h b/sysdeps/mips/linkmap.h > index 1fb9678..1e640c3 100644 > --- a/sysdeps/mips/linkmap.h > +++ b/sysdeps/mips/linkmap.h > @@ -3,4 +3,5 @@ struct link_map_machine > ElfW(Addr) plt; /* Address of .plt */ > ElfW(Word) fpabi; /* FP ABI of the object */ > unsigned int odd_spreg; /* Does the object require odd_spreg support? */ > + const Elf32_Word *mips_xlat_zero; /* .MIPS.xhash */ > }; > diff --git a/sysdeps/unix/sysv/linux/mips/ldsodefs.h b/sysdeps/unix/sysv/linux/mips/ldsodefs.h > index 8dde855..ce7b2f9 100644 > --- a/sysdeps/unix/sysv/linux/mips/ldsodefs.h > +++ b/sysdeps/unix/sysv/linux/mips/ldsodefs.h > @@ -34,7 +34,7 @@ extern void _dl_static_init (struct link_map *map); > #undef VALID_ELF_ABIVERSION > #define VALID_ELF_ABIVERSION(osabi,ver) \ > (ver == 0 \ > - || (osabi == ELFOSABI_SYSV && ver < 4) \ > + || (osabi == ELFOSABI_SYSV && ver < 6) \ > || (osabi == ELFOSABI_GNU && ver < LIBC_ABI_MAX)) > > #endif /* ldsodefs.h */ I presume the version 6 is to make room for the PT_GNU_STACK. > diff --git a/sysdeps/unix/sysv/linux/mips/libc-abis b/sysdeps/unix/sysv/linux/mips/libc-abis > index eaea558..83a667b 100644 > --- a/sysdeps/unix/sysv/linux/mips/libc-abis > +++ b/sysdeps/unix/sysv/linux/mips/libc-abis > @@ -16,3 +16,6 @@ UNIQUE > MIPS_O32_FP64 mips*-*-linux* > # Absolute (SHN_ABS) symbols working correctly. > ABSOLUTE > +# GNU-style hash table with translation table. > +MIPS_XHASH > + > Ok.
Hello, >> The corresponding binutils patch can be found here: >> https://sourceware.org/ml/binutils/2019-07/msg00098.html > If I understood correctly this patch has not been pushed upstream, neither > in fact actually reviewed. My understanding is we require upstream support > for either kernel or toolchain features before including it on glibc, although > not sure about this specific feature where it really tied with binutils > support. > > Carlos, do you think we should wait for binutils? Binutils patch hasn't been pushed upstream, but it was reviewed in [1]. We are currently waiting on FSF to process the binutils assignment for RT-RK [2]. >> @@ -955,6 +955,12 @@ _dl_setup_hash (struct link_map *map) >> map->l_gnu_buckets = hash32; >> hash32 += map->l_nbuckets; >> map->l_gnu_chain_zero = hash32 - symbias; >> + >> + /* Initialize MIPS xhash translation table. */ >> +#ifdef ELF_MACHINE_XHASH_SETUP >> + ELF_MACHINE_XHASH_SETUP (hash32, symbias, map) >> +#endif >> + >> return; >> } >> > I think it is better we move away from ifdef such kind of logic where possible > and just define ELF_MACHINE_XHASH_SETUP in the generic as an empty macro. Acknowledged. Defined as an empty macro in the newly submitted patch [3]. >> diff --git a/sysdeps/mips/linkmap.h b/sysdeps/mips/linkmap.h >> index 1fb9678..1e640c3 100644 >> --- a/sysdeps/mips/linkmap.h >> +++ b/sysdeps/mips/linkmap.h >> @@ -3,4 +3,5 @@ struct link_map_machine >> ElfW(Addr) plt; /* Address of .plt */ >> ElfW(Word) fpabi; /* FP ABI of the object */ >> unsigned int odd_spreg; /* Does the object require odd_spreg support? */ >> + const Elf32_Word *mips_xlat_zero; /* .MIPS.xhash */ >> }; >> diff --git a/sysdeps/unix/sysv/linux/mips/ldsodefs.h b/sysdeps/unix/sysv/linux/mips/ldsodefs.h >> index 8dde855..ce7b2f9 100644 >> --- a/sysdeps/unix/sysv/linux/mips/ldsodefs.h >> +++ b/sysdeps/unix/sysv/linux/mips/ldsodefs.h >> @@ -34,7 +34,7 @@ extern void _dl_static_init (struct link_map *map); >> #undef VALID_ELF_ABIVERSION >> #define VALID_ELF_ABIVERSION(osabi,ver) \ >> (ver == 0 \ >> - || (osabi == ELFOSABI_SYSV && ver < 4) \ >> + || (osabi == ELFOSABI_SYSV && ver < 6) \ >> || (osabi == ELFOSABI_GNU && ver < LIBC_ABI_MAX)) >> >> #endif /* ldsodefs.h */ > I presume the version 6 is to make room for the PT_GNU_STACK. Actually, the jump from 4 to 5 should have happened when support for absolute symbols was added to binutils [4], but it didn't, so I had to address that here. Cheers, Mihailo [1] https://sourceware.org/ml/binutils/2019-07/msg00055.html [2] https://sourceware.org/ml/binutils/2019-07/msg00215.html [3] https://sourceware.org/ml/libc-alpha/2019-07/msg00526.html [4] https://github.com/bminor/binutils-gdb/blob/bfa2a36d94d124eb7b54fd271a543047579b47ee/bfd/elfxx-mips.c#L16606
On Tue, 23 Jul 2019, Mihailo Stojanović wrote: > Binutils patch hasn't been pushed upstream, but it was reviewed in [1]. > We are currently waiting on FSF to process the binutils assignment for > RT-RK [2]. Then we should await that before considering this patch for glibc. In my view, nothing should be entered as a candidate on a release blocker list unless the changes to all other relevant packages are upstream in those components *before the start of the release freeze* (as well as the glibc patch itself being essentially ready before the release freeze). > > I presume the version 6 is to make room for the PT_GNU_STACK. > > Actually, the jump from 4 to 5 should have happened when support > for absolute symbols was added to binutils [4], but it didn't, so I had > to address that here. Bug fixes should not be mixed with new features like this. Rather, post a bug fix patch that updates the version for the absolute symbols changes *and adds a testcase to the glibc testsuite that fails without the patch* (if there's some reason it's hard to test, that needs to be adequately explained in the proposed commit message for the patch). Once that's gone through review and some version has been accepted, then a GNU hash patch building on top of it can be considered. The fix relating to absolute symbols might be considered for backporting to whatever release branches have the relevant feature but without the update to ABI versions accepted, whereas GNU hash support is a new feature clearly not appropriate for such backporting.
diff --git a/elf/dl-addr.c b/elf/dl-addr.c index 9d285d7..f504e03 100644 --- a/elf/dl-addr.c +++ b/elf/dl-addr.c @@ -20,7 +20,6 @@ #include <stddef.h> #include <ldsodefs.h> - static inline void __attribute ((always_inline)) determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, @@ -42,7 +41,7 @@ determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, ElfW(Word) strtabsize = match->l_info[DT_STRSZ]->d_un.d_val; const ElfW(Sym) *matchsym = NULL; - if (match->l_info[ADDRIDX (DT_GNU_HASH)] != NULL) + if (match->l_info[ELF_MACHINE_GNU_HASH_ADDRIDX] != NULL) { /* We look at all symbol table entries referenced by the hash table. */ @@ -57,6 +56,7 @@ determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, { /* The hash table never references local symbols so we can omit that test here. */ + symndx = ELF_MACHINE_HASH_SYMIDX(match, hasharr); if ((symtab[symndx].st_shndx != SHN_UNDEF || symtab[symndx].st_value != 0) && symtab[symndx].st_shndx != SHN_ABS @@ -65,8 +65,6 @@ determine_info (const ElfW(Addr) addr, struct link_map *match, Dl_info *info, matchsym, addr) && symtab[symndx].st_name < strtabsize) matchsym = (ElfW(Sym) *) &symtab[symndx]; - - ++symndx; } while ((*hasharr++ & 1u) == 0); } diff --git a/elf/dl-lookup.c b/elf/dl-lookup.c index e3f42a1..d717ed3 100644 --- a/elf/dl-lookup.c +++ b/elf/dl-lookup.c @@ -403,7 +403,7 @@ do_lookup_x (const char *undef_name, uint_fast32_t new_hash, do if (((*hasharr ^ new_hash) >> 1) == 0) { - symidx = hasharr - map->l_gnu_chain_zero; + symidx = ELF_MACHINE_HASH_SYMIDX (map, hasharr); sym = check_match (undef_name, ref, version, flags, type_class, &symtab[symidx], symidx, strtab, map, &versioned_sym, @@ -937,10 +937,10 @@ _dl_setup_hash (struct link_map *map) { Elf_Symndx *hash; - if (__glibc_likely (map->l_info[ADDRIDX (DT_GNU_HASH)] != NULL)) + if (__glibc_likely (map->l_info[ELF_MACHINE_GNU_HASH_ADDRIDX] != NULL)) { Elf32_Word *hash32 - = (void *) D_PTR (map, l_info[ADDRIDX (DT_GNU_HASH)]); + = (void *) D_PTR (map, l_info[ELF_MACHINE_GNU_HASH_ADDRIDX]); map->l_nbuckets = *hash32++; Elf32_Word symbias = *hash32++; Elf32_Word bitmask_nwords = *hash32++; @@ -955,6 +955,12 @@ _dl_setup_hash (struct link_map *map) map->l_gnu_buckets = hash32; hash32 += map->l_nbuckets; map->l_gnu_chain_zero = hash32 - symbias; + + /* Initialize MIPS xhash translation table. */ +#ifdef ELF_MACHINE_XHASH_SETUP + ELF_MACHINE_XHASH_SETUP (hash32, symbias, map) +#endif + return; } diff --git a/elf/elf.h b/elf/elf.h index bb13c87..b53d815 100644 --- a/elf/elf.h +++ b/elf/elf.h @@ -1715,6 +1715,7 @@ typedef struct #define SHT_MIPS_EH_REGION 0x70000027 #define SHT_MIPS_XLATE_OLD 0x70000028 #define SHT_MIPS_PDR_EXCEPTION 0x70000029 +#define SHT_MIPS_XHASH 0x7000002b /* Legal values for sh_flags field of Elf32_Shdr. */ @@ -1962,7 +1963,9 @@ typedef struct in a PIE as it stores a relative offset from the address of the tag rather than an absolute address. */ #define DT_MIPS_RLD_MAP_REL 0x70000035 -#define DT_MIPS_NUM 0x36 +/* GNU-style hash table with xlat. */ +#define DT_MIPS_XHASH 0x70000036 +#define DT_MIPS_NUM 0x37 /* Legal values for DT_MIPS_FLAGS Elf32_Dyn entry. */ diff --git a/sysdeps/generic/ldsodefs.h b/sysdeps/generic/ldsodefs.h index b1fc5c3..e95d8cd 100644 --- a/sysdeps/generic/ldsodefs.h +++ b/sysdeps/generic/ldsodefs.h @@ -47,6 +47,17 @@ __BEGIN_DECLS #define ADDRIDX(tag) (DT_NUM + DT_THISPROCNUM + DT_VERSIONTAGNUM \ + DT_EXTRANUM + DT_VALNUM + DT_ADDRTAGIDX (tag)) +/* Type of GNU hash which the machine uses. */ +#ifndef ELF_MACHINE_GNU_HASH_ADDRIDX +# define ELF_MACHINE_GNU_HASH_ADDRIDX ADDRIDX (DT_GNU_HASH) +#endif + +/* Calculate the index of a symbol in GNU_HASH. */ +#ifndef ELF_MACHINE_HASH_SYMIDX +# define ELF_MACHINE_HASH_SYMIDX(map, hasharr) \ + ((hasharr) - (map)->l_gnu_chain_zero) +#endif + /* We use this macro to refer to ELF types independent of the native wordsize. `ElfW(TYPE)' is used in place of `Elf32_TYPE' or `Elf64_TYPE'. */ #define ELFW(type) _ElfW (ELF, __ELF_NATIVE_CLASS, type) diff --git a/sysdeps/mips/ldsodefs.h b/sysdeps/mips/ldsodefs.h index f0acb02..0c6f027 100644 --- a/sysdeps/mips/ldsodefs.h +++ b/sysdeps/mips/ldsodefs.h @@ -26,6 +26,17 @@ struct La_mips_32_retval; struct La_mips_64_regs; struct La_mips_64_retval; +#define ELF_MACHINE_GNU_HASH_ADDRIDX (DT_MIPS_XHASH - DT_LOPROC + DT_NUM) + +/* Calculate the index of a symbol in MIPS_XHASH. */ +#define ELF_MACHINE_HASH_SYMIDX(map, hasharr) \ + ((map)->l_mach.mips_xlat_zero[(hasharr) - (map)->l_gnu_chain_zero]) + +/* Setup MIPS_XHASH. */ +#define ELF_MACHINE_XHASH_SETUP(hash32, symbias, map) \ + (hash32) += (map)->l_info[DT_MIPS (SYMTABNO)]->d_un.d_val - (symbias); \ + (map)->l_mach.mips_xlat_zero = (hash32) - (symbias); + #define ARCH_PLTENTER_MEMBERS \ Elf32_Addr (*mips_o32_gnu_pltenter) (Elf32_Sym *, unsigned int, \ uintptr_t *, uintptr_t *, \ diff --git a/sysdeps/mips/linkmap.h b/sysdeps/mips/linkmap.h index 1fb9678..1e640c3 100644 --- a/sysdeps/mips/linkmap.h +++ b/sysdeps/mips/linkmap.h @@ -3,4 +3,5 @@ struct link_map_machine ElfW(Addr) plt; /* Address of .plt */ ElfW(Word) fpabi; /* FP ABI of the object */ unsigned int odd_spreg; /* Does the object require odd_spreg support? */ + const Elf32_Word *mips_xlat_zero; /* .MIPS.xhash */ }; diff --git a/sysdeps/unix/sysv/linux/mips/ldsodefs.h b/sysdeps/unix/sysv/linux/mips/ldsodefs.h index 8dde855..ce7b2f9 100644 --- a/sysdeps/unix/sysv/linux/mips/ldsodefs.h +++ b/sysdeps/unix/sysv/linux/mips/ldsodefs.h @@ -34,7 +34,7 @@ extern void _dl_static_init (struct link_map *map); #undef VALID_ELF_ABIVERSION #define VALID_ELF_ABIVERSION(osabi,ver) \ (ver == 0 \ - || (osabi == ELFOSABI_SYSV && ver < 4) \ + || (osabi == ELFOSABI_SYSV && ver < 6) \ || (osabi == ELFOSABI_GNU && ver < LIBC_ABI_MAX)) #endif /* ldsodefs.h */ diff --git a/sysdeps/unix/sysv/linux/mips/libc-abis b/sysdeps/unix/sysv/linux/mips/libc-abis index eaea558..83a667b 100644 --- a/sysdeps/unix/sysv/linux/mips/libc-abis +++ b/sysdeps/unix/sysv/linux/mips/libc-abis @@ -16,3 +16,6 @@ UNIQUE MIPS_O32_FP64 mips*-*-linux* # Absolute (SHN_ABS) symbols working correctly. ABSOLUTE +# GNU-style hash table with translation table. +MIPS_XHASH +
Hello everyone, This patch is a reimplementation of [1], which was submitted back in 2015. Copyright issue has been sorted out [2] last year. It proposed a new section (.gnu.xhash) and related dynamic tag (DT_GNU_XHASH). The new section would be virtually identical to the existing .gnu.hash except for the translation table (xlat) which would contain correct MIPS .dynsym indexes corresponding to the hashvals in chains. This is because MIPS ABI imposes a different ordering of the dynyms than the one expected by the .gnu.hash section. Another addition would be a leading word which would contain the number of entries in the translation table. In this patch, the new section name and dynamic tag are changed to reflect the fact that the section should be treated as MIPS-specific (.MIPS.xhash and DT_MIPS_XHASH). This patch addresses the alignment issue as reported in [3] which is caused by the leading word added to the .MIPS.xhash section. Leading word is now removed in the corresponding binutils patch, and the number of entries in the translation table is computed using DT_MIPS_SYMTABNO dynamic tag. Since the MIPS specific dl-lookup.c file was removed since the initial patch, I opted for the definition of three new macros in the generic ldsodefs.h. ELF_MACHINE_GNU_HASH_ADDRIDX defines the index of the dynamic tag in the l_info array. ELF_MACHINE_HASH_SYMIDX is used to calculate the index of a symbol in GNU hash. On MIPS, it is defined to look up the symbol index in the translation table. ELF_MACHINE_XHASH_SETUP is defined for MIPS only. It initializes the .MIPS.xhash pointer in the link_map_machine struct. The other major change is reserving MIPS ABI version 5 for .MIPS.xhash, suggesting that the dynamic linker now supports .MIPS.xhash. This is something of which I am not sure, especially after reading [4]. I am confused on whether this ABI version is reserved for IFUNC, or it can be used for this purpose. The patch was tested by running the glibc testsuite for the three MIPS ABIs (o32, n32 and n64). The corresponding binutils patch can be found here: https://sourceware.org/ml/binutils/2019-07/msg00098.html Regards, Mihailo [1] https://sourceware.org/ml/binutils/2015-10/msg00057.html [2] https://sourceware.org/ml/binutils/2018-03/msg00025.html [3] https://sourceware.org/ml/binutils/2016-01/msg00006.html [4] https://sourceware.org/ml/libc-alpha/2016-12/msg00853.html * elf/dl-addr.c (determine_info): Calculate the symbol index using the newly defined ELF_MACHINE_HASH_SYMIDX macro. * elf/dl-lookup.c (do_lookup_x): Calculate the symbol index using the newly defined ELF_MACHINE_HASH_SYMIDX macro. (_dl_setup_hash): Initialize MIPS xhash translation table. * elf/elf.h (SHT_MIPS_XHASH): New define. (DT_MIPS_XHASH): New define. * sysdeps/generic/ldsodefs.h (ELF_MACHINE_GNU_HASH_ADDRIDX): New define. (ELF_MACHINE_HASH_SYMIDX): Ditto. *sysdeps/mips/ldsodefs.h (ELF_MACHINE_GNU_HASH_ADDRIDX): New define. (ELF_MACHINE_HASH_SYMIDX): Ditto. (ELF_MACHINE_XHASH_SETUP): Ditto. * sysdeps/mips/linkmap.h (struct link_map_machine): New member. * sysdeps/unix/sysv/linux/mips/ldsodefs.h: Increment valid ABI version. * sysdeps/unix/sysv/linux/mips/libc-abis: New ABI version. --- elf/dl-addr.c | 6 ++---- elf/dl-lookup.c | 12 +++++++++--- elf/elf.h | 5 ++++- sysdeps/generic/ldsodefs.h | 11 +++++++++++ sysdeps/mips/ldsodefs.h | 11 +++++++++++ sysdeps/mips/linkmap.h | 1 + sysdeps/unix/sysv/linux/mips/ldsodefs.h | 2 +- sysdeps/unix/sysv/linux/mips/libc-abis | 3 +++ 8 files changed, 42 insertions(+), 9 deletions(-)