This is the mail archive of the
mailing list for the Cygwin project.
Re: /proc/*/cmdline corrupted
> I've checked and on Linux (at least) xterm's command line is not
That means: unchanged ?
Or, with -sl 2000 +tb removed ?
A version (!) of xterm on Linux, which does not know +tb option, does not
> From looking at the xterm code, it would appear that the X
> libraries would have to be what is corrupting the command line. I
> didn't look into the X library (XmParseCommand).
Well, it depends whether it is xterm's intention to change the command line.
(And this would be a topic for cygwin-apps list I think.)
For example sendmail likes to do that (on Linux):
root 3051 sendmail: accepting connections
smmsp 3061 sendmail: Queue runner@00:01:00 for
root 14631 sendmail: server [22.214.171.124] cmd read
root 15254 sendmail: ./p9CDUban025571 mail3.cae3.com.: user open
entries of sendmail are not nullterminated (!).
Others have set (on Linux) a number of NULL pointers:
0000000 i n i t [ 3 ] \0 \0 \0 \0 \0 \0 \0 \0 \0
0000000 l p d W a i t i n g \0 \0 \0
Hence, XmParseCommand() would do better job if it would zero all unused
not just the first unused one.
* On the side of cygwin, the question is
1. whether /proc should show the current content of command line,
that is, what is the standard today.
(I think at least $0 should have the current value, I do not know about
The "\0 \0 \0" above suggests that the Linux's behavior = cygwin's
2. How it should interpret NULL pointers in the array, either
as "for (int i = 0; i < __argc_safe; i++)" or as
"for (char **a = __argv; *a; a++)". Both are present in pinfo.cc.
The "\0 \0 \0" above suggests that the Linux does it the first way.
(Note that the implications of "\0 \0 \0" and other observations are no more
than my speculations, that cannot replace a look at the code. :-) )
View this message in context: http://old.nabble.com/-proc-*-cmdline-corrupted-tp32639066p32665348.html
Sent from the Cygwin list mailing list archive at Nabble.com.
Problem reports: http://cygwin.com/problems.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple