ehlo,
glibc contains two versions of finction stpncpy sh-4.2$ nm --dynamic --defined-only /lib64/libc.so.6 | grep stpncpy 0000003ce1c89b00 i stpncpy 0000003ce1c89b00 i __stpncpy 0000003ce1d040a0 T __stpncpy_chk
We should use more portable version of stpncpy without prefix with underscores.
The function __stpncpy was used in initial veersion e65c65fc710fa030bfb8319efc43fcdc9ce5a26f, which was based on http://people.redhat.com/drepper/SHA-crypt.txt
patch is simple replacement "s/__stpncpy/stpncpy/"
LS
On Sat, 2014-05-17 at 19:35 +0200, Lukas Slebodnik wrote:
ehlo,
glibc contains two versions of finction stpncpy sh-4.2$ nm --dynamic --defined-only /lib64/libc.so.6 | grep stpncpy 0000003ce1c89b00 i stpncpy 0000003ce1c89b00 i __stpncpy 0000003ce1d040a0 T __stpncpy_chk
We should use more portable version of stpncpy without prefix with underscores.
The function __stpncpy was used in initial veersion e65c65fc710fa030bfb8319efc43fcdc9ce5a26f, which was based on http://people.redhat.com/drepper/SHA-crypt.txt
patch is simple replacement "s/__stpncpy/stpncpy/"
LS
Sounds like the right thing do to. I was actually suprised that string.h had both functions declared, I was expecting one to be just a #define of the other. But the libc sources indeed only have one definition that aliases to both stpncpy and __stpncpy (which is also visible from the addresses of your nm test).
ACK
On Sun, 2014-05-18 at 05:38 +0200, Jakub Hrozek wrote:
On Sat, 2014-05-17 at 19:35 +0200, Lukas Slebodnik wrote:
ehlo,
glibc contains two versions of finction stpncpy sh-4.2$ nm --dynamic --defined-only /lib64/libc.so.6 | grep stpncpy 0000003ce1c89b00 i stpncpy 0000003ce1c89b00 i __stpncpy 0000003ce1d040a0 T __stpncpy_chk
We should use more portable version of stpncpy without prefix with underscores.
The function __stpncpy was used in initial veersion e65c65fc710fa030bfb8319efc43fcdc9ce5a26f, which was based on http://people.redhat.com/drepper/SHA-crypt.txt
patch is simple replacement "s/__stpncpy/stpncpy/"
LS
Sounds like the right thing do to. I was actually suprised that string.h had both functions declared, I was expecting one to be just a #define of the other. But the libc sources indeed only have one definition that aliases to both stpncpy and __stpncpy (which is also visible from the addresses of your nm test).
ACK
master: 83e1ea9e24aabe6ade180649ecfc9c253ad19395
On Sun, May 18, 2014 at 05:44:40AM +0200, Jakub Hrozek wrote:
On Sun, 2014-05-18 at 05:38 +0200, Jakub Hrozek wrote:
On Sat, 2014-05-17 at 19:35 +0200, Lukas Slebodnik wrote:
ehlo,
glibc contains two versions of finction stpncpy sh-4.2$ nm --dynamic --defined-only /lib64/libc.so.6 | grep stpncpy 0000003ce1c89b00 i stpncpy 0000003ce1c89b00 i __stpncpy 0000003ce1d040a0 T __stpncpy_chk
We should use more portable version of stpncpy without prefix with underscores.
The function __stpncpy was used in initial veersion e65c65fc710fa030bfb8319efc43fcdc9ce5a26f, which was based on http://people.redhat.com/drepper/SHA-crypt.txt
patch is simple replacement "s/__stpncpy/stpncpy/"
LS
Sounds like the right thing do to. I was actually suprised that string.h had both functions declared, I was expecting one to be just a #define of the other. But the libc sources indeed only have one definition that aliases to both stpncpy and __stpncpy (which is also visible from the addresses of your nm test).
ACK
master: 83e1ea9e24aabe6ade180649ecfc9c253ad19395
sssd-1-11: eb774aff170b3ea671dacaf50aee5595cce1ae9a
sssd-devel@lists.fedorahosted.org