.pmc not loaded if @INC has a trailing slash

[p5pRT]

Migrated from rt.perl.org#121512 (status was 'resolved')

Searchable as RT121512$

[p5pRT]

From @schwern

Created by @schwern

If the @​INC directory where a .pm and .pmc are located has a trailing slash\, require will fail to load the .pmc. I suspect this bug was added in eb70bb4a400e88a66c7e10414a2d52b5da4cfd1f

$ pwd /Users/schwern/tmp

$ perl -I ~/tmp -wle 'use Foo; print Foo​::test()' pmc

$ perl -I ~/tmp/ -wle 'use Foo; print Foo​::test()' pm

$ cat Foo.pm package Foo;

sub test { "pm" }

=pod

=head1 NAME

Foo - this is the pm

=cut

1;

$ cat Foo.pmc package Foo;

sub test { "pmc" }

=pod

=head1 NAME

Foo - this is the pmc

=cut

1;

This bug appears in... 5.18.1 5.18.2 5.19.10

This bug does not appear in... 5.10.1 5.12.5 5.14.4 5.16.3

Perl Info

``` Flags: category=core severity=medium Site configuration information for perl 5.18.2: Configured by schwern at Thu Mar 6 14:40:23 PST 2014. Summary of my perl5 (revision 5 version 18 subversion 2) configuration: Platform: osname=darwin, osvers=13.1.0, archname=darwin-2level uname='darwin windhund.local 13.1.0 darwin kernel version 13.1.0: thu jan 16 19:40:37 pst 2014; root:xnu-2422.90.20~2release_x86_64 x86_64 i386 macbookpro8,1 darwin ' config_args='-de -Dprefix=/Users/schwern/perl5/perlbrew/perls/perl-5.18.2 -Aeval:scriptdir=/Users/schwern/perl5/perlbrew/perls/perl-5.18.2/bin' hint=recommended, useposix=true, d_sigaction=define useithreads=undef, usemultiplicity=undef useperlio=define, d_sfio=undef, uselargefiles=define, usesocks=undef use64bitint=define, use64bitall=define, uselongdouble=undef usemymalloc=n, bincompat5005=undef Compiler: cc='cc', ccflags ='-fno-common -DPERL_DARWIN -fno-strict-aliasing -pipe -fstack-protector -I/opt/local/include', optimize='-O3', cppflags='-fno-common -DPERL_DARWIN -fno-strict-aliasing -pipe -fstack-protector -I/opt/local/include' ccversion='', gccversion='4.2.1 Compatible Apple LLVM 5.0 (clang-500.2.79)', gccosandvers='' intsize=4, longsize=8, ptrsize=8, doublesize=8, byteorder=12345678 d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16 ivtype='long', ivsize=8, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8 alignbytes=8, prototype=define Linker and Libraries: ld='env MACOSX_DEPLOYMENT_TARGET=10.3 cc', ldflags =' -fstack-protector -L/opt/local/lib' libpth=/opt/local/lib /usr/lib libs=-lgdbm -ldbm -ldl -lm -lutil -lc perllibs=-ldl -lm -lutil -lc libc=, so=dylib, useshrplib=false, libperl=libperl.a gnulibc_version='' Dynamic Linking: dlsrc=dl_dlopen.xs, dlext=bundle, d_dlsymun=undef, ccdlflags=' ' cccdlflags=' ', lddlflags=' -bundle -undefined dynamic_lookup -L/opt/local/lib -fstack-protector' Locally applied patches: @INC for perl 5.18.2: /Users/schwern/perl5/perlbrew/perls/perl-5.18.2/lib/site_perl/5.18.2/darwin-2level /Users/schwern/perl5/perlbrew/perls/perl-5.18.2/lib/site_perl/5.18.2 /Users/schwern/perl5/perlbrew/perls/perl-5.18.2/lib/5.18.2/darwin-2level /Users/schwern/perl5/perlbrew/perls/perl-5.18.2/lib/5.18.2 . Environment for perl 5.18.2: DYLD_LIBRARY_PATH (unset) HOME=/Users/schwern LANG=en_US.UTF-8 LANGUAGE (unset) LD_LIBRARY_PATH (unset) LOGDIR (unset) PATH=/Users/schwern/perl5/perlbrew/bin:/Users/schwern/perl5/perlbrew/perls/perl-5.18.2/bin:/Users/schwern/bin:/opt/local/libexec/gnubin:/opt/local/bin:/opt/local/sbin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/local/MacGPG2/bin PERLBREW_BASHRC_VERSION=0.67 PERLBREW_HOME=/Users/schwern/.perlbrew PERLBREW_MANPATH=/Users/schwern/perl5/perlbrew/perls/perl-5.18.2/man PERLBREW_PATH=/Users/schwern/perl5/perlbrew/bin:/Users/schwern/perl5/perlbrew/perls/perl-5.18.2/bin PERLBREW_PERL=perl-5.18.2 PERLBREW_ROOT=/Users/schwern/perl5/perlbrew PERLBREW_VERSION=0.67 PERL_BADLANG (unset) SHELL=/bin/bash ```

[p5pRT]

From @iabyn

On Wed\, Mar 26\, 2014 at 03​:23​:06PM -0700\, Michael G Schwern wrote​:

If the @​INC directory where a .pm and .pmc are located has a trailing slash\, require will fail to load the .pmc. I suspect this bug was added in eb70bb4a400e88a66c7e10414a2d52b5da4cfd1f

A bisect comes up with this​:

commit 6b0bdd7f2041803dc3ec72b53d28052705861967 Author​: Matthew Horsfall (alh) \wolfsage@​gmail\.com AuthorDate​: Thu Dec 27 10​:38​:08 2012 -0500 Commit​: Father Chrysostomos \sprout@​cpan\.org CommitDate​: Sun Feb 10 12​:19​:15 2013 -0800

  RT-116192 - If a directory in @​INC already has a trailing '/'\, don't add another.

-- Nothing ventured\, nothing lost.

[p5pRT]

The RT System itself - Status changed from 'new' to 'open'

[p5pRT]

From @wolfsage

On Sat\, Mar 29\, 2014 at 3​:28 PM\, Dave Mitchell \davem@​iabyn\.com wrote​:

On Wed\, Mar 26\, 2014 at 03​:23​:06PM -0700\, Michael G Schwern wrote​:

If the @​INC directory where a .pm and .pmc are located has a trailing slash\, require will fail to load the .pmc. I suspect this bug was added in eb70bb4a400e88a66c7e10414a2d52b5da4cfd1f

A bisect comes up with this​:

commit 6b0bdd7f2041803dc3ec72b53d28052705861967 Author​: Matthew Horsfall (alh) \wolfsage@​gmail\.com AuthorDate​: Thu Dec 27 10​:38​:08 2012 -0500 Commit​: Father Chrysostomos \sprout@​cpan\.org CommitDate​: Sun Feb 10 12​:19​:15 2013 -0800

RT\-116192 \- If a directory in @​INC already has a trailing '/'\, don't add another\.

My apologies! Patch attached.

The original code always assumed a '/' would be appended\, so it set the length of the resulting SV to be +1.

With my earlier patch\, this wasn't the case if the SV already ended in a '/'\, so later checks on the SV that were based off the length of the SV would be wrong.

New tests included.

-- Matthew Horsfall (alh)

[p5pRT]

From @wolfsage

0001-RT-121512-Allow-I-dir-with-trailing-slash-to-find-.p.patch

```diff From 4f86a010d75c228240961c2e480bae9a51653237 Mon Sep 17 00:00:00 2001 From: "Matthew Horsfall (alh)" Date: Mon, 31 Mar 2014 08:25:33 -0400 Subject: [PATCH] RT-121512 - Allow -I/dir/ with trailing slash to find .pmc files. 6b0bdd7f2041803dc3ec72b53d28052705861967 updated -I to not add a trailing '/' if one was already present, but failed to update the length of the resulting SV to account for the lack of another character. This caused later checks based off of that length to fail (in this case, seeing if the last 3 characters of the string are .pm). --- MANIFEST | 2 ++ pp_ctl.c | 5 ++++- t/run/flib/t2.pm | 5 +++++ t/run/flib/t2.pmc | 5 +++++ t/run/switchM.t | 10 +++++++++- 5 files changed, 25 insertions(+), 2 deletions(-) create mode 100644 t/run/flib/t2.pm create mode 100644 t/run/flib/t2.pmc diff --git a/MANIFEST b/MANIFEST index 70565d5..0794152 100644 --- a/MANIFEST +++ b/MANIFEST @@ -5444,6 +5444,8 @@ t/run/dtrace.pl For dtrace.t t/run/dtrace.t Test for DTrace probes t/run/exit.t Test perl's exit status. t/run/flib/broken.pm Bad .pm file for switchM.t +t/run/flib/t2.pm Test for .pmcs with -I/dir/ +t/run/flib/t2.pmc Test for .pmcs with -I/dir/ t/run/fresh_perl.t Tests that require a fresh perl. t/run/locale.t Tests related to locale handling t/run/mad.t Test vs MAD environment diff --git a/pp_ctl.c b/pp_ctl.c index 3c643d7..1352d38 100644 --- a/pp_ctl.c +++ b/pp_ctl.c @@ -3954,6 +3954,7 @@ PP(pp_require) if (path_searchable) { const char *dir; STRLEN dirlen; + U8 ext_len = 1; if (SvOK(dirsv)) { dir = SvPV_nomg_const(dirsv, dirlen); @@ -3999,13 +4000,15 @@ PP(pp_require) /* Avoid '//' */ if (!dirlen || *(tmp-1) != '/') { *tmp++ = '/'; + } else { + ext_len = 0; } /* name came from an SV, so it will have a '\0' at the end that we can copy as part of this memcpy(). */ memcpy(tmp, name, len + 1); - SvCUR_set(namesv, dirlen + len + 1); + SvCUR_set(namesv, dirlen + len + ext_len); SvPOK_on(namesv); } # endif diff --git a/t/run/flib/t2.pm b/t/run/flib/t2.pm new file mode 100644 index 0000000..172a14e --- /dev/null +++ b/t/run/flib/t2.pm @@ -0,0 +1,5 @@ +package t2; + +sub id { "t2pm" } + +1; diff --git a/t/run/flib/t2.pmc b/t/run/flib/t2.pmc new file mode 100644 index 0000000..e3894bc --- /dev/null +++ b/t/run/flib/t2.pmc @@ -0,0 +1,5 @@ +package t2; + +sub id { "t2pmc" } + +1; diff --git a/t/run/switchM.t b/t/run/switchM.t index 72e8908..6a75100 100644 --- a/t/run/switchM.t +++ b/t/run/switchM.t @@ -8,7 +8,7 @@ use strict; require './test.pl'; -plan(2); +plan(4); like(runperl(switches => ['-Irun/flib', '-Mbroken'], stderr => 1), qr/^Global symbol "\$x" requires explicit package name at run\/flib\/broken.pm line 6\./, @@ -17,3 +17,11 @@ like(runperl(switches => ['-Irun/flib', '-Mbroken'], stderr => 1), like(runperl(switches => ['-Irun/flib/', '-Mbroken'], stderr => 1), qr/^Global symbol "\$x" requires explicit package name at run\/flib\/broken.pm line 6\./, "Ensure -Irun/flib/ produces correct filename in warnings"); + +like(runperl(switches => ['-Irun/flib', '-Mt2', '-e "print t2::id()"'], stderr => 1), + qr/^t2pmc$/, + "Ensure -Irun/flib loads pmc"); + +like(runperl(switches => ['-Irun/flib/', '-Mt2', '-e "print t2::id()"'], stderr => 1), + qr/^t2pmc$/, + "Ensure -Irun/flib/ loads pmc"); -- 1.7.9.5 ```

[p5pRT]

From @tonycoz

On Mon Mar 31 05​:32​:55 2014\, alh wrote​:

RT-116192 - If a directory in @​INC already has a trailing '/'\, don't add another.

My apologies! Patch attached.

The original code always assumed a '/' would be appended\, so it set the length of the resulting SV to be +1.

With my earlier patch\, this wasn't the case if the SV already ended in a '/'\, so later checks on the SV that were based off the length of the SV would be wrong.

New tests included.

I think this patch or something like it should be applied to blead and backported.

The only issue I have with the patch itself is it could be simpler\, instead of adding extra variable\, the if could simply ++dirlen when it bumps tmp.

Nothing else within the scope depends on dirlen being the length of the original @​INC entry.

Tony

[p5pRT]

From @wolfsage

On Mon\, Mar 31\, 2014 at 7​:28 PM\, Tony Cook via RT \perlbug\-followup@​perl\.org wrote​:

I think this patch or something like it should be applied to blead and backported.

The only issue I have with the patch itself is it could be simpler\, instead of adding extra variable\, the if could simply ++dirlen when it bumps tmp.

Nothing else within the scope depends on dirlen being the length of the original @​INC entry.

Thanks for the review Tony. Attached a version with your suggested changes.

Side question (related to both of these patches) - is it bad that in the case of dirs that already end with '/'\, we're calling SvGROW with one more byte than we end up using? Or is there nothing wrong with SvGROW to an arbitrary length and SvCUR_set to a smaller size?

Thanks\,

-- Matthew Horsfall (alh)

[p5pRT]

From @wolfsage

0001-RT-121512-Allow-I-dir-with-trailing-slash-to-find-.p.patch

```diff From 3277d1b90c2bf865b14f5bf27134f6c28b9d607f Mon Sep 17 00:00:00 2001 From: "Matthew Horsfall (alh)" Date: Mon, 31 Mar 2014 08:25:33 -0400 Subject: [PATCH] RT-121512 - Allow -I/dir/ with trailing slash to find .pmc files. 6b0bdd7f2041803dc3ec72b53d28052705861967 updated -I to not add a trailing '/' if one was already present, but failed to update the length of the resulting SV to account for the lack of another character. This caused later checks based off of that length to fail (in this case, seeing if the last 3 characters of the string are .pm). --- MANIFEST | 2 ++ pp_ctl.c | 3 +++ t/run/flib/t2.pm | 5 +++++ t/run/flib/t2.pmc | 5 +++++ t/run/switchM.t | 10 +++++++++- 5 files changed, 24 insertions(+), 1 deletion(-) create mode 100644 t/run/flib/t2.pm create mode 100644 t/run/flib/t2.pmc diff --git a/MANIFEST b/MANIFEST index 70565d5..0794152 100644 --- a/MANIFEST +++ b/MANIFEST @@ -5444,6 +5444,8 @@ t/run/dtrace.pl For dtrace.t t/run/dtrace.t Test for DTrace probes t/run/exit.t Test perl's exit status. t/run/flib/broken.pm Bad .pm file for switchM.t +t/run/flib/t2.pm Test for .pmcs with -I/dir/ +t/run/flib/t2.pmc Test for .pmcs with -I/dir/ t/run/fresh_perl.t Tests that require a fresh perl. t/run/locale.t Tests related to locale handling t/run/mad.t Test vs MAD environment diff --git a/pp_ctl.c b/pp_ctl.c index 3c643d7..e13e450 100644 --- a/pp_ctl.c +++ b/pp_ctl.c @@ -3999,6 +3999,9 @@ PP(pp_require) /* Avoid '//' */ if (!dirlen || *(tmp-1) != '/') { *tmp++ = '/'; + } else { + /* So SvCUR_set reports the correct length below */ + dirlen--; } /* name came from an SV, so it will have a '\0' at the diff --git a/t/run/flib/t2.pm b/t/run/flib/t2.pm new file mode 100644 index 0000000..172a14e --- /dev/null +++ b/t/run/flib/t2.pm @@ -0,0 +1,5 @@ +package t2; + +sub id { "t2pm" } + +1; diff --git a/t/run/flib/t2.pmc b/t/run/flib/t2.pmc new file mode 100644 index 0000000..e3894bc --- /dev/null +++ b/t/run/flib/t2.pmc @@ -0,0 +1,5 @@ +package t2; + +sub id { "t2pmc" } + +1; diff --git a/t/run/switchM.t b/t/run/switchM.t index 72e8908..6a75100 100644 --- a/t/run/switchM.t +++ b/t/run/switchM.t @@ -8,7 +8,7 @@ use strict; require './test.pl'; -plan(2); +plan(4); like(runperl(switches => ['-Irun/flib', '-Mbroken'], stderr => 1), qr/^Global symbol "\$x" requires explicit package name at run\/flib\/broken.pm line 6\./, @@ -17,3 +17,11 @@ like(runperl(switches => ['-Irun/flib', '-Mbroken'], stderr => 1), like(runperl(switches => ['-Irun/flib/', '-Mbroken'], stderr => 1), qr/^Global symbol "\$x" requires explicit package name at run\/flib\/broken.pm line 6\./, "Ensure -Irun/flib/ produces correct filename in warnings"); + +like(runperl(switches => ['-Irun/flib', '-Mt2', '-e "print t2::id()"'], stderr => 1), + qr/^t2pmc$/, + "Ensure -Irun/flib loads pmc"); + +like(runperl(switches => ['-Irun/flib/', '-Mt2', '-e "print t2::id()"'], stderr => 1), + qr/^t2pmc$/, + "Ensure -Irun/flib/ loads pmc"); -- 1.7.9.5 ```

[p5pRT]

From @tonycoz

On Tue Apr 01 09​:42​:26 2014\, alh wrote​:

Side question (related to both of these patches) - is it bad that in the case of dirs that already end with '/'\, we're calling SvGROW with one more byte than we end up using? Or is there nothing wrong with SvGROW to an arbitrary length and SvCUR_set to a smaller size?

That isn't a problem\, SvGROW() will often allocate a bit more space anyway.

pp_sysread() provides one example where we grow to a length where perhaps not all that space is used.

Tony

[p5pRT]

From @wolfsage

On Tue Apr 01 21​:12​:13 2014\, tonyc wrote​:

On Tue Apr 01 09​:42​:26 2014\, alh wrote​:

Side question (related to both of these patches) - is it bad that in the case of dirs that already end with '/'\, we're calling SvGROW with one more byte than we end up using? Or is there nothing wrong with SvGROW to an arbitrary length and SvCUR_set to a smaller size?

That isn't a problem\, SvGROW() will often allocate a bit more space anyway.

pp_sysread() provides one example where we grow to a length where perhaps not all that space is used.

Tony

Alright\, thanks. I've pushed this to blead.

Closing.

-- Matthew Horsfall (alh)

-- -- Matthew Horsfall (alh)

[p5pRT]

@wolfsage - Status changed from 'open' to 'resolved'

[p5pRT]

From @wolfsage

Relevant commit​: 9fdd5a7ac74817cfaab65f7284f98ea66faca324

-- Matthew Horsfall (alh)

[p5pRT]

From [Unknown Contact. See original ticket]

Relevant commit​: 9fdd5a7ac74817cfaab65f7284f98ea66faca324

-- Matthew Horsfall (alh)