You are viewing a plain text version of this content. The canonical link for it is here.
Posted to modperl@perl.apache.org by Stas Bekman <st...@stason.org> on 2007/04/03 00:52:50 UTC
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Stephane, you are correct in your observations that original STDIN and
STDOUT filehandles are not preserved when perlio CGI mode is used.
This is because how Perl works. Consider the following perl program, run
from the command line (not mod_perl!):
#!/usr/bin/perl
use strict;
use warnings;
warn sprintf "%-11s %d\n", "STDIN", fileno(STDIN);
open STDIN_SAVED, "<&STDIN" or die "Can't dup STDIN: $!";
warn sprintf "%-11s %d\n", "STDIN_SAVED", fileno(STDIN_SAVED);
close STDIN;
open DUMMY, ">/dev/null" or die $!;
#open DUMMY, "</dev/random" or die $!;
warn sprintf "%-11s %d\n", "DUMMY", fileno(DUMMY);
open STDIN, "<&STDIN_SAVED" or die "Can't dup STDIN_SAVED: $!";
warn sprintf "%-11s %d\n", "STDIN", fileno(STDIN);
close STDIN_SAVED;
% perl-5.8.8-ithreads /tmp/test1
STDIN 0
STDIN_SAVED 3
Filehandle STDIN reopened as DUMMY only for output at /tmp/xxx line 14.
DUMMY 0
STDIN 4
as you can see DUMMY grabs the available fd 0. (not talking about the
problem that perl handles fd=0 as special, but only warns about it, were
you to use it to open "</dev/random" it'd have been quiet
Now obviously the solution in your case is to grab fd before anybody
else does and release it just before the original STDIN is restored.
Here is the adjusted program:
#!/usr/bin/perl
use strict;
use warnings;
warn sprintf "%-11s %d\n", "STDIN", fileno(STDIN);
# new!
open STDIN_SAVED, "<&STDIN" or die "Can't dup STDIN: $!";
warn sprintf "%-11s %d\n", "STDIN_SAVED", fileno(STDIN_SAVED);
close STDIN;
open SAVE_FD, "</dev/null" or die "Can't dup STDIN: $!";
warn sprintf "%-11s %d\n", "SAVE_FD", fileno(SAVE_FD);
open DUMMY, ">/dev/null" or die $!;
#open DUMMY, "</dev/random" or die $!;
warn sprintf "%-11s %d\n", "DUMMY", fileno(DUMMY);
# new!
close SAVE_FD;
open STDIN, "<&STDIN_SAVED" or die "Can't dup STDIN_SAVED: $!";
warn sprintf "%-11s %d\n", "STDIN", fileno(STDIN);
close STDIN_SAVED;
% perl-5.8.8-ithreads /tmp/test2
STDIN 0
STDIN_SAVED 3
SAVE_FD 0
DUMMY 4
STDIN 0
which is what you want.
But which file to open as a place holder? In this example I use
/dev/null and it works as you wish. AFAIK, windows doesn't have
/dev/null, unless you run on cygwin.
I thought to simply get the fd number and open ala fdopen,
[ open SAVE_FD, "<&=STDIN" ]
as you can see in the following example:
#!/usr/bin/perl
use strict;
use warnings;
warn sprintf "%-11s %d\n", "STDIN", fileno(STDIN);
# opening STDIN's fd directly, no dup!
open SAVE_FD, "<&=STDIN" or die "Can't dup STDIN: $!";
warn sprintf "%-11s %d\n", "SAVE_FD", fileno(SAVE_FD);
open STDIN_SAVED, "<&STDIN" or die "Can't dup STDIN: $!";
warn sprintf "%-11s %d\n", "STDIN_SAVED", fileno(STDIN_SAVED);
close STDIN;
warn sprintf "%-11s %d\n", "SAVE_FD", fileno(SAVE_FD);
open DUMMY, ">/dev/null" or die $!;
#open DUMMY, "</dev/random" or die $!;
warn sprintf "%-11s %d\n", "DUMMY", fileno(DUMMY);
# new!
close SAVE_FD;
open STDIN, "<&STDIN_SAVED" or die "Can't dup STDIN_SAVED: $!";
warn sprintf "%-11s %d\n", "STDIN", fileno(STDIN);
close STDIN_SAVED;
% perl-5.8.8-ithreads /tmp/test3
STDIN 0
SAVE_FD 0
STDIN_SAVED 3
SAVE_FD 0
Filehandle STDIN reopened as DUMMY only for output at /tmp/xxx1 line 18.
DUMMY 4
STDIN 0
as you can see that it works almost well:
+ the fd is restored correctly
- perl is not happy that we open a non "<" fh when STDIN is closed.
[this sounds like a bug in perl, but perlio is a very complicated thing,
so I won't be surprised if that's by design]
You can choose not to use perlio, by re-compiling perl with -Uuseperlio
and then rebuilding perl. I think it should do the trick as well, though
I didn't test it.
Here is a proof of concept patch for modperl_io.c, that does what you
want keeping perlio enabled (with a lot of debug statements). Though it
needs more work to permit re-entrance [hardcoded GENX2|4 symbols], which
will be quite tricky, since it requires internal API change, or some
trickery. [the patch is attached as well, if things get wrapped up]
Index: src/modules/perl/modperl_io.c
===================================================================
--- src/modules/perl/modperl_io.c (revision 524915)
+++ src/modules/perl/modperl_io.c (working copy)
@@ -108,14 +108,20 @@
{
dHANDLE("STDIN");
int status;
- GV *handle_save = (GV*)Nullsv;
+ GV *handle_save = (GV*)Nullsv;
+ GV *handle_save_fd = (GV*)Nullsv;
+
SV *sv = sv_newmortal();
+ fprintf(stderr, "STDIN orig: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle))));
+
MP_TRACE_o(MP_FUNC, "start");
/* if STDIN is open, dup it, to be restored at the end of response */
if (handle && SvTYPE(handle) == SVt_PVGV &&
IoTYPE(GvIO(handle)) != IoTYPE_CLOSED) {
+
handle_save = gv_fetchpv(Perl_form(aTHX_
"Apache2::RequestIO::_GEN_%ld",
(long)PL_gensym++),
@@ -132,6 +138,21 @@
* have file descriptors, so STDIN must be closed before it can
* be reopened */
do_close(handle, TRUE);
+
+ /* now grab the just released fd, normally 0 */
+ handle_save_fd = gv_fetchpv("GENX2", TRUE, SVt_PVIO);
+
+ /* open my $oldout, "<&=0" or die "Can't save STDIN's fd: $!"; */
+ status = do_open(handle_save_fd, "</dev/null", 10, FALSE,
+ O_RDONLY, 0, Nullfp);
+ if (status == 0) {
+ Perl_croak(aTHX_ "Failed to save STDIN's fd: %" SVf,
+ get_sv("!", TRUE));
+ }
+
+ fprintf(stderr, "saved STDIN fd: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle_save_fd))));
+
}
sv_setref_pv(sv, "Apache2::RequestRec", (void*)r);
@@ -142,6 +163,9 @@
}
MP_TRACE_o(MP_FUNC, "end\n");
+
+ fprintf(stderr, "STDIN dupped: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle))));
return handle_save;
}
@@ -152,8 +176,12 @@
dHANDLE("STDOUT");
int status;
GV *handle_save = (GV*)Nullsv;
+ GV *handle_save_fd = (GV*)Nullsv;
SV *sv = sv_newmortal();
+ fprintf(stderr, "STDOUT orig: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle))));
+
MP_TRACE_o(MP_FUNC, "start");
/* if STDOUT is open, dup it, to be restored at the end of response */
@@ -175,6 +203,21 @@
* have file descriptors, so STDOUT must be closed before it can
* be reopened */
do_close(handle, TRUE);
+
+
+ /* now grab the just released fd, normally 0 */
+ handle_save_fd = gv_fetchpv("GENX4", TRUE, SVt_PVIO);
+
+ /* open my $oldout, "<&=0" or die "Can't save STDIN's fd: $!"; */
+ status = do_open(handle_save_fd, ">/dev/null", 10, FALSE,
+ O_RDONLY, 0, Nullfp);
+ if (status == 0) {
+ Perl_croak(aTHX_ "Failed to save STDOUT's fd: %" SVf,
+ get_sv("!", TRUE));
+ }
+
+ fprintf(stderr, "saved STDOUT fd: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle_save_fd))));
}
sv_setref_pv(sv, "Apache2::RequestRec", (void*)r);
@@ -190,14 +233,17 @@
* overridden? */
IoFLUSH_off(handle); /* STDOUT's $|=0 */
+ fprintf(stderr, "STDOUT dupped: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle))));
+
return handle_save;
-
}
MP_INLINE void modperl_io_perlio_restore_stdin(pTHX_ GV *handle)
{
GV *handle_orig = gv_fetchpv("STDIN", FALSE, SVt_PVIO);
-
+ GV *handle_save_fd = gv_fetchpv("GENX2", TRUE, SVt_PVIO);
+
MP_TRACE_o(MP_FUNC, "start");
/* close the overriding filehandle */
@@ -212,6 +258,8 @@
MP_TRACE_o(MP_FUNC, "restoring STDIN");
+ do_close(handle_save_fd, FALSE);
+
if (do_open9(handle_orig, "<&", 2, FALSE,
O_RDONLY, 0, Nullfp, (SV*)handle, 1) == 0) {
err = get_sv("!", TRUE);
@@ -226,13 +274,16 @@
}
}
+ fprintf(stderr, "STDIN restored: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle_orig))));
+
MP_TRACE_o(MP_FUNC, "end\n");
}
MP_INLINE void modperl_io_perlio_restore_stdout(pTHX_ GV *handle)
{
GV *handle_orig = gv_fetchpv("STDOUT", FALSE, SVt_PVIO);
-
+ GV *handle_save_fd = gv_fetchpv("GENX4", TRUE, SVt_PVIO);
MP_TRACE_o(MP_FUNC, "start");
/* since closing unflushed STDOUT may trigger a subrequest
@@ -259,6 +310,8 @@
MP_TRACE_o(MP_FUNC, "restoring STDOUT");
+ do_close(handle_save_fd, FALSE);
+
if (do_open9(handle_orig, ">&", 2, FALSE,
O_WRONLY, 0, Nullfp, (SV*)handle, 1) == 0) {
err = get_sv("!", TRUE);
@@ -273,5 +326,8 @@
}
}
+ fprintf(stderr, "STDOUT restored: %d\n",
+ (int)PerlIO_fileno(IoIFP(GvIOn(handle_orig))));
+
MP_TRACE_o(MP_FUNC, "end\n");
}
--
_____________________________________________________________
Stas Bekman mailto:stas@stason.org http://stason.org/
http://www.linkedin.com/in/stasbekman http://stasosphere.com/
The "Practical mod_perl" book http://modperlbook.org/
http://stason.org/photos/gallery/ http://healingcloud.com
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Posted by Stas Bekman <st...@stason.org>.
Stephane Chazelas wrote:
> On Mon, Apr 02, 2007 at 03:52:50PM -0700, Stas Bekman wrote:
>> Stephane, you are correct in your observations that original STDIN and
>> STDOUT filehandles are not preserved when perlio CGI mode is used.
>>
>> This is because how Perl works. Consider the following perl program, run from the command line (not mod_perl!):
>>
>
> Hi Stas, that's not what I think happens though.
>
> I suspected as well that the responsible code was in
> src/modules/perl/modperl_io.c
>
> And it does IIUC:
>
> open STDIN_SAVED, "<&STDIN" or die "Can't dup STDIN: $!";
> close STDIN;
> open STDIN, "<:apache"
>
> and same for STDOUT.
>
> And it should work, except that the second open, which should
> resolve to a dup2 or dup system call doesn't seem to do anything
> file-descriptor-wise. It rather looks like is assignes the STDIN
> perl handle to the Socket, so that STDIN no longer has the fd 0
> (as demonstrated by my test of executing a lsof command within a
> CGI script), but something more like 16, which was returned by
> the accept() of the incoming HTTP connection.
That's correct, Stephane. The :Apache perlio layer does not use a
socket. It uses $r instead, which is already interfacing an existing
non-STD socket. That's why the :Apache perlio layer doesn't take over
the just released fd.
There is a problem with using a socket with :Apache perlio layer. If we
do so Perl will attempt to use that socket. I suppose it's still
possible to work around that, you are more than welcome to change the
:Apache perlio layer to dup(2). It can be found in
modperl_io_apache.[ch] files. modperl_io.c invokes it via open
"<:Apache". You can find the documentation of perlio layers in the
perliol manpage.
>> handle_save = gv_fetchpv(Perl_form(aTHX_
>> "Apache2::RequestIO::_GEN_%ld",
[...]
>> + /* now grab the just released fd, normally 0 */
>> + handle_save_fd = gv_fetchpv("GENX2", TRUE, SVt_PVIO);
>> +
>> + /* open my $oldout, "<&=0" or die "Can't save STDIN's fd: $!"; */
>> + status = do_open(handle_save_fd, "</dev/null", 10, FALSE,
>> + O_RDONLY, 0, Nullfp);
> [...]
>
> You don't want that IMO, you want the fd 0 and 1 to point to the
> socket so that the CGI has both its stdin and stdout pointing to
> the socket, because that's how unmodified CGI scripts do.
>
> system("echo foo");
>
> Should output "foo\n" to the page returned by the CGI. and echo
> foo does a write(1, "foo\n", 4);
>
> I must confess I tried to follow in the perl code what that
> do_open9 call that was supposed to dup the socket into
> stdin/stdout was doing but didn't get very far. I tried to
> replace those perl functions to real close() and dup()s but it
> didn't help either.
Yes, do_openn (do_open9 is just a wrapper) is hairy. I spent hours
deciphering it. I hope the above explanation makes the issue with lack
of dup() clear.
--
_____________________________________________________________
Stas Bekman mailto:stas@stason.org http://stason.org/
http://www.linkedin.com/in/stasbekman http://stasosphere.com/
The "Practical mod_perl" book http://modperlbook.org/
http://stason.org/photos/gallery/ http://healingcloud.com
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Posted by Stephane Chazelas <st...@artesyncp.com>.
On Mon, Apr 02, 2007 at 03:52:50PM -0700, Stas Bekman wrote:
> Stephane, you are correct in your observations that original STDIN and
> STDOUT filehandles are not preserved when perlio CGI mode is used.
>
> This is because how Perl works. Consider the following perl program, run from the command line (not mod_perl!):
>
Hi Stas, that's not what I think happens though.
I suspected as well that the responsible code was in
src/modules/perl/modperl_io.c
And it does IIUC:
open STDIN_SAVED, "<&STDIN" or die "Can't dup STDIN: $!";
close STDIN;
open STDIN, "<:apache"
and same for STDOUT.
And it should work, except that the second open, which should
resolve to a dup2 or dup system call doesn't seem to do anything
file-descriptor-wise. It rather looks like is assignes the STDIN
perl handle to the Socket, so that STDIN no longer has the fd 0
(as demonstrated by my test of executing a lsof command within a
CGI script), but something more like 16, which was returned by
the accept() of the incoming HTTP connection.
[...]
> + fprintf(stderr, "STDIN orig: %d\n",
> + (int)PerlIO_fileno(IoIFP(GvIOn(handle))));
> +
> MP_TRACE_o(MP_FUNC, "start");
>
> /* if STDIN is open, dup it, to be restored at the end of response */
> if (handle && SvTYPE(handle) == SVt_PVGV &&
> IoTYPE(GvIO(handle)) != IoTYPE_CLOSED) {
> +
> handle_save = gv_fetchpv(Perl_form(aTHX_
> "Apache2::RequestIO::_GEN_%ld",
> (long)PL_gensym++),
> @@ -132,6 +138,21 @@
> * have file descriptors, so STDIN must be closed before it can
> * be reopened */
> do_close(handle, TRUE);
> +
> + /* now grab the just released fd, normally 0 */
> + handle_save_fd = gv_fetchpv("GENX2", TRUE, SVt_PVIO);
> +
> + /* open my $oldout, "<&=0" or die "Can't save STDIN's fd: $!"; */
> + status = do_open(handle_save_fd, "</dev/null", 10, FALSE,
> + O_RDONLY, 0, Nullfp);
[...]
You don't want that IMO, you want the fd 0 and 1 to point to the
socket so that the CGI has both its stdin and stdout pointing to
the socket, because that's how unmodified CGI scripts do.
system("echo foo");
Should output "foo\n" to the page returned by the CGI. and echo
foo does a write(1, "foo\n", 4);
I must confess I tried to follow in the perl code what that
do_open9 call that was supposed to dup the socket into
stdin/stdout was doing but didn't get very far. I tried to
replace those perl functions to real close() and dup()s but it
didn't help either.
--
Stephane
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Posted by Stas Bekman <st...@stason.org>.
William A. Rowe, Jr. wrote:
> Stas Bekman wrote:
>> William A. Rowe, Jr. wrote:
>>> Stas Bekman wrote:
>>>> But which file to open as a place holder? In this example I use
>>>> /dev/null and it works as you wish. AFAIK, windows doesn't have
>>>> /dev/null, unless you run on cygwin.
>>> Of course, it does, but it has the classic MS-DOS name of "NUL"
>>> present in the context of any directory (so \dev\nul works if you
>>> have a dev directory.)
>> I guess this can be ifdef'ed then.
>>
>> Bill, so /dev/null and NUL should cover all grounds Apache runs on, or
>> are there other odd cases? And if so do you have a better idea for a
>> file to open that will work everywhere? [also notice that it has to be
>> rw - and if not, we need two files then, one for < and another for >,
>> otherwise open will fail.
>
> AFAIK - dev@apr might be an interesting collection of folks to ask.
Hopefully someone here has an equivalent or a better knowledge.
Thanks for your help, Bill
--
_____________________________________________________________
Stas Bekman mailto:stas@stason.org http://stason.org/
http://www.linkedin.com/in/stasbekman http://stasosphere.com/
The "Practical mod_perl" book http://modperlbook.org/
http://stason.org/photos/gallery/ http://healingcloud.com
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Stas Bekman wrote:
> William A. Rowe, Jr. wrote:
>> Stas Bekman wrote:
>>> But which file to open as a place holder? In this example I use
>>> /dev/null and it works as you wish. AFAIK, windows doesn't have
>>> /dev/null, unless you run on cygwin.
>>
>> Of course, it does, but it has the classic MS-DOS name of "NUL"
>> present in the context of any directory (so \dev\nul works if you
>> have a dev directory.)
>
> I guess this can be ifdef'ed then.
>
> Bill, so /dev/null and NUL should cover all grounds Apache runs on, or
> are there other odd cases? And if so do you have a better idea for a
> file to open that will work everywhere? [also notice that it has to be
> rw - and if not, we need two files then, one for < and another for >,
> otherwise open will fail.
AFAIK - dev@apr might be an interesting collection of folks to ask.
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Posted by Stas Bekman <st...@stason.org>.
William A. Rowe, Jr. wrote:
> Stas Bekman wrote:
>> But which file to open as a place holder? In this example I use
>> /dev/null and it works as you wish. AFAIK, windows doesn't have
>> /dev/null, unless you run on cygwin.
>
> Of course, it does, but it has the classic MS-DOS name of "NUL"
> present in the context of any directory (so \dev\nul works if you
> have a dev directory.)
I guess this can be ifdef'ed then.
Bill, so /dev/null and NUL should cover all grounds Apache runs on, or
are there other odd cases? And if so do you have a better idea for a
file to open that will work everywhere? [also notice that it has to be
rw - and if not, we need two files then, one for < and another for >,
otherwise open will fail.
Thanks Bill
--
_____________________________________________________________
Stas Bekman mailto:stas@stason.org http://stason.org/
http://www.linkedin.com/in/stasbekman http://stasosphere.com/
The "Practical mod_perl" book http://modperlbook.org/
http://stason.org/photos/gallery/ http://healingcloud.com
Re: Fw: Re: mod_perl closes apache's stdin and/or stdout
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Stas Bekman wrote:
>
> But which file to open as a place holder? In this example I use
> /dev/null and it works as you wish. AFAIK, windows doesn't have
> /dev/null, unless you run on cygwin.
Of course, it does, but it has the classic MS-DOS name of "NUL"
present in the context of any directory (so \dev\nul works if you
have a dev directory.)