On Sunday 09 October 2005 03:49, Steve Underwood wrote:
> Robert Diemer wrote:
> >Hello,
> >thank you for your answer, Steve. Unfortunatly your version from liHIL
> > don't work, too. There is the same error.
> >If I make a "strace gdbproxy msp430" (libHIL in 32bit version) I get the
> >following output:
> >...
> >open("/dev/parport0", O_RDWR) = 3
> >ioctl(3, PPCLAIM, 0xffffcf28) = 0
> >ioctl(3, PPDATADIR, 0xffffce34) = -1 EINVAL (Invalid argument)
> >dup(2) = 4
> >fcntl64(4, F_GETFL) = 0x8002 (flags
> > O_RDWR|O_LARGEFILE) brk(0) = 0x8103000
> >brk(0x8124000) = 0x8124000
> >fstat64(0x4, 0xffffcc6c) = 0
> >...
> >
> >On a 32bit computer the syscall "ioctl(3, PPDATADIR, 0xffffce34)" works,
> > the result is 0 and not -1.
> >
> >
> >And if I use a short test-program with libHIL, both in 64bit mode, it
> > works too.
> >Therefore I think the problem is the syscall.
> >
> >Now, I don't know how to isolate the problem. I thought that the problem
> > was the conversion of the pointer in the ioctl-syscall to unsigned long.
> >
> >Has anybody an idea?
> >Steve, can you tell me please which library you use with gdbproxy?
> >
> >My ldd-output is
> > linux-gate.so.1 => (0xffffe000)
> > libHIL.so => /usr/local/msp430//lib32/libHIL.so (0x5556d000)
> > libc.so.6 => /lib32/libc.so.6 (0x55591000)
> > /lib/ld-linux.so.2 (0x55555000)
> >These are all 32bit libraries.
>
> I guess the conversion of the unsigned long is the source of the
> problem. However, it works out OK on other x86_64 machines. If it
> didn't, 32 bit compatibility wouldn't work. ioctl calls are used by most
> programs, though the ones to ppdev may be rare. My machine is running
> Linux 2.6.13, but note that the Fedora versions of the kernels always
> contain various patches. I used to run 2.6.12 on the same machine,
> without problems.
>
> My ldd output is:
> linux-gate.so.1 => (0xffffe000)
> libHIL.so => /usr/lib/libHIL.so (0xf7fc6000)
> libc.so.6 => /lib/libc.so.6 (0x00a33000)
> /lib/ld-linux.so.2 (0x00a15000)
>
> Regards,
> Steve
>
Thanks for your answer. I had the same thoughts, but I have not found a
solution yet. I will compare my kernel (2.6.12) with the fedory ones. Perhaps
I find the answer...Regards, Robert -- ------------------------------------------------------------------------ Dipl.-Ing. Robert Diemer http://www.rcs.ei.tum.de Institute for Real-Time Computer Systems (RCS) fon +49-89-289-23731 Technische Universitaet Muenchen, D-80290 Muenchen fax +49-89-289-23555
pgpPvQehHZRod.pgp
Description: PGP signature
