The construct certain vintages of GCC (the one I have trouble
with is 3.2.3) have trouble with looks like the following:
static struct st a;
static struct st *p = &a;
struct st { int foo; };
static void init(void) { a.foo = 0; }
The problem disappears if we move the struct declaration up to
let the compiler know the shape of the struct before the first
definition uses it, like this:
struct st { int foo; }; /* this has been moved up */
static struct st a;
static struct st *p = &a;
static void init(void) { a.foo = 0; }
"If the shell is compiled with -DJOBS, this is all fine -- find wasn't
stopped (it was killed), so it correctly uses WTERMSIG instead of WSTOPSIG.
However, if the shell _isn't_ compiled with -DJOBS (which it isn't in d-i),
only WSTOPSIG is used, which extracts the high byte instead of the low
byte from the status code. Since the status code is 13 (SIGPIPE), "st"
suddenly gets the value 0, which is equivalent to SIGEXIT. Thus, ash prints
out "EXIT" on find's exit."
unless it had #!/bin/sh in the first line
"It correctly locates the script, tries to execute it via execve which
fails. After that it tries to hand it over to /bin/sh which fails too,
since ash
- neither provides the absolute pathname to /bin/sh
- nor tries to lookup the script via PATH if called as "sh script"
"
Last patch have synced form Manuel Nova III xxreadtoken() function,
corrected (C) form dash debian/copyright, removed my small mistake
with IFS_BROKEN (thanks by Herbert), and synced cmdedit.c from
current CVS (removed libc5 support, your email correction, my (C) year
corertion).
Hello all,
This patch adds more "Help" text to the config system. Almost
all applets now have a help entry. Also, I cleaned up the spacing of
the existing text so that things are consistent. This patch is against
this morning's CVS.
Thomas Cameron
CEI Systems, Inc.
modified Kbuild system I put into uClibc. With this, there should be no more
need to modify Rules.mak since I've moved all the interesting options into the
config system. I think I've got everything updated, but you never know, I may
have made some mistakes, so watch closely.
-Erik
config system
- added a new config option to allow persistant history lists. This is
currently only used by ash, but the calls ({load,save}_history) could
be added to the other shells as well.
HP iPAQ. Sometimes we get a SIGWINCH and want to print the prompt, although
the prompt is not yet initialised (0) .. so just don't print a prompt
as long as the prompt string is 0
#49: I found one memory overflow and memory leak in "ln" applet.
Last patch reduced also 54 bytes. ;)
#50: I found bug in loginutils/Makefile.in.
New patch have also new function to libbb and
aplied this to applets and other cosmetic changes.
When alias support is not configured, ash believes that command parameters
that look like dd's "if=/dev/zero" are requests to set a temporary
environment variable whilst dd is running, even though it appears after the
command name. This is caused by the re-use of the checkalias global variable
to indicate when both alias checking and environment variable checking. The
failure to reset this flag is due to the reset action being performed only
inside the feature check CHECK_ASH_ALIAS. Hence ash works as expected when
aliases are configured in, and fails when not.
Example script using 'date' with different settings of TZ:
# TZ=Europe/London
# export TZ
# date
Thu May 30 17:18:49 BST 2002
# TZ=America/New_York date
Thu May 30 12:19:10 EDT 2002
# date
Thu May 30 17:19:12 BST 2002
# date TZ=America/New_York
Thu May 30 12:19:30 EDT 2002 <----- wrong, should be BST time (or error!)
# date
Thu May 30 17:19:35 BST 2002
Attached is a patch against revision 1.52 of ash.c which moves the checks so
that checkalias is updated regardless of whether CONFIG_ASH_ALIAS is set.
With this patch applied, the command shown above which should generate an
error does generate an error.
I have tested this patch with the 'dd' command too and that now works
correctly.
the busybox development tree. This eliminates the use of recursive make, and
once again allows us to run 'make' in a subdirectory with the expected result.
And things are now much faster too. Greatly improved IMHO...
-Erik
This diff does 2 things:
1) removes an unnecessary function. saves 64 bytes on i386
2) allows you to disable checking of mail (actually, it's now disabled
by default). this would be a nice CML1 option, but for now it's a
#(define|undef) in the C file like the other internal ash options.
this saves an additional 352 bytes if you leave mail disabled.
so that the table implementations of copynode, calcsize, and cmdtxt are
used. Hopefully this will result in wider testing. ;-) If it breaks,
just unset the appropriate defines.
and copynode() table implementations. Commit the fix but keep them disabled
until others check them out. Uncomment "//#define CMDTXT_TABLE",
"//#define CALCSIZE_TABLE", and "//#define COPYNODE_TABLE" to try them out.
Saves over 600 bytes on i386.
math suport, cleaner math syntax error checking, moves redundant signal
string tables (from kill and ash) into libbb and provides a few
cleanups elsewhere.
was written by Aaron Lehmann <aaronl@vitelus.com> for busybox. This
patch makes a few trivial changes to Aaron's code so that it can be
used (in theory) by the other shells as well...
-Erik
1) ping cleanup (compile fix from this patch already applied).
2) traceroute call not spare ntohl() now (and reduce size);
3) Fix for functions not declared static in insmod, ash, vi and mount.
4) a more simple API cmdedit :))
5) adds "stopped jobs" warning to ash on Ctrl-D and fixes "ignoreeof" option
6) reduce exporting library function index->strchr (traceroute), bzero->memset (syslogd)
handles all the basic stuff (for, case/esac, while, if/then/else), and
is very small (adds just 38k on x86). It is not as rigorously correct
about Bourne semantics as bash, but for most things it is quite
workable. There is still some work to be done to further shrink it (it
has its own globbing functions instead of using the libc ones, for
example), but it is quite usable as is.
-Erik
together by vodz, but uses newer sources, has the removed features
commented out instead of simply deleted (so they could be re-enabled)
and the builtins all work. This adds 72k.
-Erik
just be a simple command line interpreter with basic pipe, redirect, and job
control. For all the more fancy things, people should use hush or ash.
-Erik
Fixes the interaction between if/then/else/fi syntax and variables.
I planned to do it right from the beginning, but my implementation
was buggy. Also adds the relevant test cases. Also adds some old
Matt Kraai variable test cases that got left out somehow.
> I rewrite *local_variable* function in hush.c with:
> 1) remove many memory leaks
> 2) add support read_only protect (require write builtin function for set this,
> I write this special for variable HUSH_VERION=0.01)
> 3) commad read set only local variable now
> 4) remove many error messages if "set unset export" not defined variable
> (bash syntax not put and set error code). Hmm, if I set result to -1, you hush
> called waitpid and returned with error "no waitpid" ( i not found place this
> error).
> 5) destroy error in new version check xgetcwd()==NULL and set "(unknow)" -
> this have error: crashe in next call `pwd`, but xgetcwd(not null) called
> free(arg).
> 6) next add integraion with libbb
Valdimir's patch missed two cases of local variable handling
FOO=bar
export FOO=baz
unset FOO
and
export FOO=bar
FOO=baz
which were working before, so I fixed those two cases.
$ a=b foo
should be handled correctly.
$ a=b
is parsed OK, but the actual variable setting is not
yet written. Except for some weird exceptions related
to quoting rules, this code passes (matches ash behavior)
all the tests I threw at it.
If someone now writes set_local_var(), and updates lookup_param()
to match, we can claim success!
- Larry
Update some comments. Generate partial placeholders for the missing
builtins. Write builtin_umask. Properly treat exec without arguments
as a means to open/close files within the running script. Implement
"4<&-" that encodes for file descriptor closure.
Minor improvements. Something is still broken with running
scripts via "hush filename". All the following are now handled
acceptably (matches ash, not bash).
if true; then echo foo1; fi
if
true; then echo foo2; fi
if true; false; then echo bar; else echo foo3; fi
if true || false; then echo foo4; fi
- Larry
It should recover more smoothly from syntax errors, and it now
has a decent guess when the reserved word construct is over
(or not) to control execution and prompting. I took out all the
redundant standalone test copies of libbb routines, but left in a
hook so I can include those for my testing. I'll post that include
file on my web site.
- Larry
April 25, 2001 snapshot, adjusted a bit by me so it has cmdedit support.
This checkin also removes sh.c. In the future sh.c will be a symlink to
your shell of choice. For now, this symlink will default to pointing to
lash.c (as in the past). If you change the symlink to point to hush.c,
then thats what you will get. This symlink business is a temporary situation,
which will be cleaned up Real Soon Now(tm).
-Erik
