[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20251112171740.GK1827@brightrain.aerifal.cx>
Date: Wed, 12 Nov 2025 12:17:40 -0500
From: Rich Felker <dalias@...c.org>
To: Tomáš Hulata <hulata@...art.tech>
Cc: musl@...ts.openwall.com
Subject: Re: [PATCH v3] vdso: add support for GNU hash tables
On Thu, Nov 06, 2025 at 10:59:45PM +0100, Tomáš Hulata wrote:
> Hi Rich,
> just reminding with this simple patch.
> It would be great to get it merged.
> Thanks
Thanks for the ping! I was probably initially planning to hold off on
this til a new release cycle, but I don't see any way it could break.
It surely won't break on any kernel that has the standard ELF hash
table available in the vdso like it should. If nobody has objections
(I'd appreciate a second set of eyes from anyone willing to read it) I
think this is ok to go ahead and apply.
Rich
> On Sun, Aug 24, 2025 at 11:25 AM Tomáš Hulata <hulata@...art.tech> wrote:
> >
> > It would be great to get it included, but it's up to you. I'm going to
> > use a fork till it is not released.
> > Thanks
> >
> > Best Regards
> > Tomáš Hulata
> > [ SysArt / Pixel Federation ]
> >
> > On Sun, Aug 24, 2025 at 4:26 AM Rich Felker <dalias@...c.org> wrote:
> > >
> > > On Sat, Aug 23, 2025 at 12:54:04AM +0200, tombo wrote:
> > > > Hi Rich, thanks for the review. Hopefully this version addresses all the points.
> > > >
> > > > ---
> > > > src/internal/vdso.c | 28 ++++++++++++++++++++++++++--
> > > > 1 file changed, 26 insertions(+), 2 deletions(-)
> > > >
> > > > diff --git a/src/internal/vdso.c b/src/internal/vdso.c
> > > > index d46d3228..2b6e1ae2 100644
> > > > --- a/src/internal/vdso.c
> > > > +++ b/src/internal/vdso.c
> > > > @@ -40,6 +40,24 @@ static int checkver(Verdef *def, int vsym, const char *vername, char *strings)
> > > > #define OK_TYPES (1<<STT_NOTYPE | 1<<STT_OBJECT | 1<<STT_FUNC | 1<<STT_COMMON)
> > > > #define OK_BINDS (1<<STB_GLOBAL | 1<<STB_WEAK | 1<<STB_GNU_UNIQUE)
> > > >
> > > > +static size_t count_syms_gnu(uint32_t *gh)
> > > > +{
> > > > + size_t nsym, i;
> > > > + uint32_t *buckets = gh + 4 + (gh[2]*sizeof(size_t)/4);
> > > > + uint32_t *hashval;
> > > > + for (i = nsym = 0; i < gh[0]; i++) {
> > > > + if (buckets[i] > nsym)
> > > > + nsym = buckets[i];
> > > > + }
> > > > + if (nsym) {
> > > > + hashval = buckets + gh[0] + (nsym - gh[1]);
> > > > + do nsym++;
> > > > + while (!(*hashval++ & 1));
> > > > + }
> > > > + return nsym;
> > > > +}
> > > > +
> > > > +
> > > > void *__vdsosym(const char *vername, const char *name)
> > > > {
> > > > size_t i;
> > > > @@ -60,6 +78,7 @@ void *__vdsosym(const char *vername, const char *name)
> > > > char *strings = 0;
> > > > Sym *syms = 0;
> > > > Elf_Symndx *hashtab = 0;
> > > > + uint32_t *ghashtab = 0;
> > > > uint16_t *versym = 0;
> > > > Verdef *verdef = 0;
> > > >
> > > > @@ -69,15 +88,20 @@ void *__vdsosym(const char *vername, const char *name)
> > > > case DT_STRTAB: strings = p; break;
> > > > case DT_SYMTAB: syms = p; break;
> > > > case DT_HASH: hashtab = p; break;
> > > > + case DT_GNU_HASH: ghashtab = p; break;
> > > > case DT_VERSYM: versym = p; break;
> > > > case DT_VERDEF: verdef = p; break;
> > > > }
> > > > }
> > > >
> > > > - if (!strings || !syms || !hashtab) return 0;
> > > > + if (!strings || !syms) return 0;
> > > > if (!verdef) versym = 0;
> > > > + size_t nsym = 0;
> > > > +
> > > > + if (hashtab) nsym = hashtab[1];
> > > > + else if (ghashtab) nsym = count_syms_gnu(ghashtab);
> > > >
> > > > - for (i=0; i<hashtab[1]; i++) {
> > > > + for (i=0; i<nsym; i++) {
> > > > if (!(1<<(syms[i].st_info&0xf) & OK_TYPES)) continue;
> > > > if (!(1<<(syms[i].st_info>>4) & OK_BINDS)) continue;
> > > > if (!syms[i].st_shndx) continue;
> > > > --
> > > > 2.46.0
> > >
> > > Thanks! I don't see anything obviously wrong with this. I'll take a
> > > more detailed look over it soon. Are you hoping to get it included in
> > > this release cycle?
> > >
> > > Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
