Re: Bug#164766: Bug #164766 C3 no cmov fix does not work
At 25 Jan 2003 17:10:06 +0100,
Christoph Martin wrote:
> [1 <text/plain (quoted-printable)>]
> Am Sam, 2003-01-25 um 15.47 schrieb Scott Ashcroft:
> > Even with the latest version of libc (2.3.1-10) and libssl0.9.7
> > (0.9.7-4) the fix does not appear to work.
> >
> > If I understand the fix correctly, on processors without cmov ld.so
> > should select the non-i686 library but it still looks like it still
> > select the i686/cmov version.
> >
> > # cat /proc/cpuinfo
> > processor : 0
> > vendor_id : CentaurHauls
> > cpu family : 6
> > model : 7
> > model name : VIA Ezra
> > stepping : 8
> > cpu MHz : 800.058
> > cache size : 64 KB
> > fdiv_bug : no
> > hlt_bug : no
> > f00f_bug : no
> > coma_bug : no
> > fpu : yes
> > fpu_exception : yes
> > cpuid level : 1
> > wp : yes
> > flags : fpu de tsc msr cx8 mtrr pge mmx 3dnow
> > bogomips : 1576.96
> >
> > # ldd `which ssh`
> > libutil.so.1 => /lib/libutil.so.1 (0x40016000)
> > libz.so.1 => /lib/libz.so.1 (0x4001a000)
> > libnsl.so.1 => /lib/libnsl.so.1 (0x40028000)
> > libcrypto.so.0.9.7 => /usr/lib/i686/cmov/libcrypto.so.0.9.7 (0x4003b000)
> > libc.so.6 => /lib/libc.so.6 (0x4012b000)
> > libdl.so.2 => /lib/libdl.so.2 (0x4023b000)
> > /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
> >
> > # ldconfig -p | grep libcrypto.so.0.9.7
> > libcrypto.so.0.9.7 (libc6, hwcap: 0x8000000008000) =>
> > /usr/lib/i686/cmov/libcrypto.so.0.9.7
> > libcrypto.so.0.9.7 (libc6, hwcap: 0x4000000000000) =>
> > /usr/lib/i586/libcrypto.so.0.9.7
> > libcrypto.so.0.9.7 (libc6, hwcap: 0x2000000000000) =>
> > /usr/lib/i486/libcrypto.so.0.9.7
> > libcrypto.so.0.9.7 (libc6) => /usr/lib/libcrypto.so.0.9.7
> >
> > Bug in ld.so (libc6), incorrect building of libssl or just my lack of
> > understanding?
>
> I would say, it is a bug in libc6. If you don't have cmov the linker
> should not load libraries in i686/cmov
-10 introduced including i686/cmov to search dir. However it seems it
lacks excluding this dir on C3. Yes, it's libc6 problem.
I investigate it for -11.
I wonder "i586/mmx" has really worked for a long time...
Regards,
-- gotom
Reply to: