Hi,
Steven said it is okay to forward this to the mailing list.
Any suggestions?
Werner
----- Forwarded message from "Steven M. Bellovin" <smb@research.att.com> -----
To: Werner Koch <wk@isil.d.shuttle.de>
Subject: Re: gpg
Date: Sat, 16 Jan 1999 08:35:49 -0500
From: "Steven M. Bellovin" <smb@research.att.com>
In message <19990116091343.B19544@isil.d.shuttle.de>, Werner Koch writes:
>"Steven M. Bellovin" <smb@research.att.com> writes:
>
>> I haven't tried running it yet; however, gpg compiles on BSD/OS 4.0 if
>> and only
>> if I specify --disable-asm, and doesn't compile at all on BSD/OS 3.1
>
>What CPU, can you (or somebody else) look at the assembler code for
>this CPU - maybe it is only a problem with the underscore prepended to
>the function names?
I believe it's the version of the assembler that's being used:
S654$ as -v
GNU assembler version 2.8.1 (bsd/os), using BFD version 2.8.1
^?S655$ gcc -v
gcc version 2.7.2.1
gcc -DHAVE_CONFIG_H -I. -I. -I.. -I../include -g -Wall -O2 -c mpih-mul1.S
/tmp/cc008707.s: Assembler messages:
/tmp/cc008707.s:117: Error: Alignment not a power of 2
/tmp/cc008707.s:135: Error: Alignment not a power of 2
*** Error code 1
>
>> of strange problems in twofish.c
>
>Try to replace the -O1 by none and tell me again please.
>
>
> werner
>
>
Here was the problem in twofish.c -- there were many more lines like this:
`echo gcc -DHAVE_CONFIG_H -I. -I. -I.. -I../include -I../intl -g -Wall
-DIS_MOD
ULE -shared -fPIC -lc -o twofish ./twofish.c | sed -e 's/-O[0-9]*/ /' `
/usr/include/stdio.h:357: warning: `__sputc' defined but not used
twofish.c:2442:invalid character '[' in first operand
twofish.c:2461:invalid character '@' in first operand
twofish.c:2464:invalid character '@' in first operand
twofish.c:2467:invalid character '@' in first operand
twofish.c:2472:invalid character '@' in first operand
twofish.c:2476:invalid character '@' in first operand
twofish.c:2479:invalid character '@' in first operand
...
twofish.c:29594:invalid character '@' in first operand
twofish.c:29599:invalid character '@' in first operand
twofish.c:29606:invalid character '@' in first operand
twofish.c:29662:invalid character '[' in first operand
twofish.c:29689:invalid character '@' in first operand
twofish.c:29699:invalid character '@' in first operand
twofish.c:29727:invalid character '@' in first operand
twofish.c:29740:invalid character '@' in first operand
s651$ gcc -v
gcc version 2.7.2.1
s652$ as -v
GNU assembler version 1.38
I confess I don't understand the structure here that involves linking
twofish.c -- if it's something to do with elf-type dynamic linking, it
wouldn't work anyway on BSD/OS 3.1.
I ran gcc with -E; it produced a rational-looking file, without 29000 lines
or #line entries. It seems to be the -fPIC that does it; if I delete that,
it compiles just fine.
I should note that I'm not that concerned about 3.1; I'm getting a new machine
in a few weeks and I'll install 4.0 on it, rather than upgrading. But you
might want to note the problem on your Web page.
----- End forwarded message -----
--
Koch Softwaresysteme / "The GNU Privacy Guard" is an OpenPGP system:
Remscheider Str. 22 / http://www.gnupg.org (ftp.guug.de:/pub/gcrypt)
D-40215 Düsseldorf /
Germany / Fingerprint for DSA key <werner.koch@guug.de>:
+49 211 3180023 / ecaf 7590 eb34 43b5 c7cf 3acb 6c7e e1b8 621c c013
Steven said it is okay to forward this to the mailing list.
Any suggestions?
Werner
----- Forwarded message from "Steven M. Bellovin" <smb@research.att.com> -----
To: Werner Koch <wk@isil.d.shuttle.de>
Subject: Re: gpg
Date: Sat, 16 Jan 1999 08:35:49 -0500
From: "Steven M. Bellovin" <smb@research.att.com>
In message <19990116091343.B19544@isil.d.shuttle.de>, Werner Koch writes:
>"Steven M. Bellovin" <smb@research.att.com> writes:
>
>> I haven't tried running it yet; however, gpg compiles on BSD/OS 4.0 if
>> and only
>> if I specify --disable-asm, and doesn't compile at all on BSD/OS 3.1
>
>What CPU, can you (or somebody else) look at the assembler code for
>this CPU - maybe it is only a problem with the underscore prepended to
>the function names?
I believe it's the version of the assembler that's being used:
S654$ as -v
GNU assembler version 2.8.1 (bsd/os), using BFD version 2.8.1
^?S655$ gcc -v
gcc version 2.7.2.1
gcc -DHAVE_CONFIG_H -I. -I. -I.. -I../include -g -Wall -O2 -c mpih-mul1.S
/tmp/cc008707.s: Assembler messages:
/tmp/cc008707.s:117: Error: Alignment not a power of 2
/tmp/cc008707.s:135: Error: Alignment not a power of 2
*** Error code 1
>
>> of strange problems in twofish.c
>
>Try to replace the -O1 by none and tell me again please.
>
>
> werner
>
>
Here was the problem in twofish.c -- there were many more lines like this:
`echo gcc -DHAVE_CONFIG_H -I. -I. -I.. -I../include -I../intl -g -Wall
-DIS_MOD
ULE -shared -fPIC -lc -o twofish ./twofish.c | sed -e 's/-O[0-9]*/ /' `
/usr/include/stdio.h:357: warning: `__sputc' defined but not used
twofish.c:2442:invalid character '[' in first operand
twofish.c:2461:invalid character '@' in first operand
twofish.c:2464:invalid character '@' in first operand
twofish.c:2467:invalid character '@' in first operand
twofish.c:2472:invalid character '@' in first operand
twofish.c:2476:invalid character '@' in first operand
twofish.c:2479:invalid character '@' in first operand
...
twofish.c:29594:invalid character '@' in first operand
twofish.c:29599:invalid character '@' in first operand
twofish.c:29606:invalid character '@' in first operand
twofish.c:29662:invalid character '[' in first operand
twofish.c:29689:invalid character '@' in first operand
twofish.c:29699:invalid character '@' in first operand
twofish.c:29727:invalid character '@' in first operand
twofish.c:29740:invalid character '@' in first operand
s651$ gcc -v
gcc version 2.7.2.1
s652$ as -v
GNU assembler version 1.38
I confess I don't understand the structure here that involves linking
twofish.c -- if it's something to do with elf-type dynamic linking, it
wouldn't work anyway on BSD/OS 3.1.
I ran gcc with -E; it produced a rational-looking file, without 29000 lines
or #line entries. It seems to be the -fPIC that does it; if I delete that,
it compiles just fine.
I should note that I'm not that concerned about 3.1; I'm getting a new machine
in a few weeks and I'll install 4.0 on it, rather than upgrading. But you
might want to note the problem on your Web page.
----- End forwarded message -----
--
Koch Softwaresysteme / "The GNU Privacy Guard" is an OpenPGP system:
Remscheider Str. 22 / http://www.gnupg.org (ftp.guug.de:/pub/gcrypt)
D-40215 Düsseldorf /
Germany / Fingerprint for DSA key <werner.koch@guug.de>:
+49 211 3180023 / ecaf 7590 eb34 43b5 c7cf 3acb 6c7e e1b8 621c c013