This is the mail archive of the libc-help@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Version mismatch libpthread/libthread_db [SOLVED]


> This isn't the usual name for a native gdb, but as long as it's native
> and you're on the target then you're OK.
> [...]
>   
> You are starting to confuse people here. Which gdb, cross-gdb running
> on the host, or native-gdb examining the running gdbserver on the
> target?
>   

That may be because I'm confused myself about the meaning of "native" in
this context. From the start: I use the above mentioned cross-compiled
mipsel-linux-gnu-gdb which runs on my host/workstation and debugs the
remote target's binaries via gdbserver on the target machine. Is that
what you asked me to do? If not, what is the native gdb (BTW, using the
hosts default gdb fails on "target remote ..." with an unknown
architecture error)?

> In general solib-absolute-prefix is only used if you have a multiple
> sets of libraries on the host and you want the cross-gdb to select a
> different set to debug with.
>   

Well, in the end that's what seemed to be causing my problems, if I
understood you and Ryan correctly? Namely that the cross-gdb was
searching for the shared libraries in the hosts root dir instead of the
target's libs that have been used by cross-gcc/ld, finding a libc and
libpthread that didn't match the target files and therefor the version
check and even stepping over calls into libc failed. Currently it looks
like setting solib-absolute-prefix, or better yet using the correct
--with-sysroot in configure (which somehow was left out from cross-gdb
configuration, as I figured out this afternoon) solves my problem entirely.
So I would like to apologize for the fuss, which in the end seems to be
a GDB configuration problem. I simply wasn't able to connect the version
mismatch error showing up on the target with a possible problem in the
cross-gdb setup on my workstation, and was looking for the solution in
libc's configuration. (Who would have thought that apparently the
cross-gdb sends the version of the libpthread found locally to
gdbserver, so it is compared and reclaimed on the target?) Thanks to
Carlos, Ryan and Mike for taking the time, providing me useful hints and
increasing my know-how!

Martin


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]