Hi Dave,
Well, as much as I know about pyinstaller, it uses the system libraries and packages the .so files in the app, sort of like Mozilla does with browsers. So, since there are only PowerPC libraries available, I was thinking it was a header problem. I did spend some time and managed to figure out the header format enough to get it to be the correct 'endian' and also to report that it was a 32 bit PowerPC app. If you check the headers for something like /bin/cp, that's what they say also.
Hmm... that gives me an idea. I wonder if the 32 bit libraries are being included?
I think I did mention that running the python script for the app runs just fine ?
I will put that on my list of things to do. You will see a post from me on something else that I would like to see resolved about the Xorg port.
Thanks for your help.
-- Jim
On Mar 23, 2014, at 2:59 PM, David Gosselin wrote:
Hi Jim,
System V should be ok as that’s just an API at this point. If the
original executable were compiled for 386, then changing the header won’t fix it
for PowerPC. Do you have the GNU compilers installed? I think that
you’ll have to rebuild this package from scratch on your system.
Or is this truly a PowerPC executable and the ELF header has been
corrupted? Either way, I think rebuilding from source will be the best
path forward.
Dave
Sent: Friday, March 21, 2014 12:35 PM
Subject: Re: Python pyinstaller produces bad binary
file
Sure
does!
I was reading the specs for binary headers for 64 bit PowerPC and it looks
like this is a 32 bit Intel file and
the address that is jumped to to execute the program is all 0's ?
Check me on this as I haven't looked at this kind of stuff since college
(long time ago). Looks like
Magic Number is OK
32 bit file
Little Endian
Original ELF version
System V ??????? Should be Linux, I think ?
ABI is 1 (Not sure if that's good or not)
Executable
PowerPC
Original ELF version (again)
Jump to 0x9680
Thanks,
--Jim
On Mar 20, 2014, at 5:28 PM, David Gosselin wrote:
No problem, Jim, glad to help. It looks like 386 binaries landed on your
PPC system from the pyinstaller.
Thanks for quick reply, Dave. Here is the
output from using 'file' on the original file: XOlog0.8.9: ELF 32-bit
LSB executable, Intel 80386, version 1 (SYSV), dynamically linked (uses
shared libs), for GNU/Linux 2.6.18,
BuildID[sha1]=0x0b7ec69824a1ff621c8e697a32d12f1cf42274c6, not
stripped and from the modified binary: XOlog0.8.9mod: ELF
32-bit MSB executable, Intel 80386 - invalid byte order, (SYSV), statically
linked, stripped Obviously, that was a disaster. --
Jim On 03/19/2014 05:53 PM, dave@appleside.org wrote:
What is the output of the ‘file’ command on the original unmodified
binary and also on the modified binary? Can you send along the
results for both? Thanks!
Dave
Hi, I have Debian 7.4 working fairly
well on a 20 inch iMac G5 . Python seems to work OK, except
that pyinstaller produces a binary file that the shell refuses to
execute. Looking at the file with beav (hex editor) I find that the
header differs from other files that do execute. I have
'fixed' these bytes, but still no joy on executing the
file. Its been a long time since I hacked on file headers,etc, so
any thoughts or help would be appreciated. Here is what the
shell says: bash: ./XOlog0.8.9: cannot execute binary
file And a dump of the file's header from
beav.
0: 7F 45 4C 46 01 01 01 00 00 00 00 00 00 00 00 00
.ELF............
10: 02 00 03 00 01 00 00 00 80 96 04 08 34 00 00 00
............4...
20: 60 6C 01 00 00 00 00 00 34 00 20 00 07 00 28 00 `l......4.
...(. 30: 25 00
22 00 06 00 00 00 34 00 00 00 34 80 04 08
%.".....4...4... Thanks, Jim --
To UNSUBSCRIBE, email to debian-powerpc-REQUEST@lists.debian.orgwith
a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.orgArchive:
[🔎] 5329CDA2.1050807@gmail.com">https://lists.debian.org/ [🔎] 5329CDA2.1050807@gmail.com
|