RE: [PHP-CVS] com php-src: Fix #71882: Negative ftruncate() on php://memory exhausts memory: NEWS ext/standard/file.c
ext/standard/tests/file/bug71882.phpt
| From: | Anatol Belski | Date: | Wed, 31 Aug 2016 11:05:57 +0000 |
| Subject: | RE: [PHP-CVS] com php-src: Fix #71882: Negative ftruncate() on php://memory exhausts memory: NEWS ext/standard/file.c ext/standard/tests/file/bug71882.phpt |
||
| References: | 1 2 3 4 5 | Groups: | php.cvs |
| Request: | Send a blank email to php-cvs+get-94187@lists.php.net to get a copy of this message | ||
> -----Original Message-----
> From: Christoph M. Becker [mailto:cmbecker69@gmx.de]
> Sent: Wednesday, August 31, 2016 12:30 PM
> To: Anatol Belski <anatol.php@belski.net>; 'Matteo Beccati'
> <php@beccati.com>; php-cvs@lists.php.net
> Cc: 'Ferenc Kovacs' <tyrael@php.net>; 'Davey Shafik'
> <davey@php.net>
> Subject: Re: [PHP-CVS] com php-src: Fix #71882: Negative ftruncate() on
> php://memory exhausts memory: NEWS ext/standard/file.c
> ext/standard/tests/file/bug71882.phpt
>
> Hi Anatol!
>
> On 31.08.2016 at 12:04, Anatol Belski wrote:
> >
> >> -----Original Message-----
> >> From: Christoph M. Becker [mailto:cmbecker69@gmx.de]
> >> Sent: Wednesday, August 31, 2016 11:45 AM
> >> To: Matteo Beccati <php@beccati.com>; Anatol Belski
> >> <anatol.php@belski.net>; php-cvs@lists.php.net
> >> Cc: 'Ferenc Kovacs' <tyrael@php.net>; Davey Shafik
> >> <davey@php.net>
> >> Subject: Re: [PHP-CVS] com php-src: Fix #71882: Negative ftruncate()
> >> on php://memory exhausts memory: NEWS ext/standard/file.c
> >> ext/standard/tests/file/bug71882.phpt
> >>
> >> On 31.08.2016 at 09:47, Matteo Beccati wrote:
> >>
> >>> negative size was tested in 4 existing phpt files, that are now
> >>> failing due to the new warning. That seems a bit of a BC break to me.
> >>
> >> Okay, it might be better not to throw the warning (for BC), but in my
> >> opinion the (zend)_long to size_t conversion has to be properly
> >> handled generally, not only for memory streams.
> >
> > What is your suggestion? There's some general approach with
> Zend/zend_range_check.h, which still could be extended for +-/*, etc. But there
> are also many special cases where some individual solution is needed.
>
> My suggestion is:
>
> if (size < 0) {
> /* php_error_docref(NULL TSRMLS_CC, E_WARNING, "Negative size is not
> supported"); */
> RETURN_FALSE;
> }
>
> And to uncomment the php_error_docref() call for master (maybe also 7.1).
>
ACK. Probably 7.1+ can live with the warning then, doesn't sound like biggie. It's an
error case, and the C runtime warning will be delivered later down the function anyway, say with
plain files.
> >> I suggest to let ftruncate() silently fail (returning false) for
> >> negative sizes in PHP
> >> 5.6 - 7.1, but to add a warning for master.
> >>
> >> Any objections?
> >
> > IMO the warning is useful, but the important part is the negative value check.
> Either way, the tests need to pass.
>
> Of course, I will fix the tests, but how to do this depends on whether
> ftruncate() will raise E_WARNING or not, so this should be decided first. :-)
>
> >>>
> >>> https://revive.beccati.com/bamboo/browse/PHP-SRC/latest
> >>>
> >>> In fact ftruncate_variation4.phpt is dubbed "usage variations -
> >>> truncate file to negative size". This is a reminder that we all
> >>> should try to run the test suite before pushing changes, at least
> >>> for the subset of functionalities that are affected by the commit ;)
> >>
> >> Running the tests is one thing, but ext/standard/tests/file/ has
> >> about
> >> 90 failing test for me anyway. That's probably because I'm running
> >> in a Vagrant Debian box on a Windows host in an NTFS folder. Running
> >> on Windows I still have nearly 20 failing test, what's probably
> >> caused by running an admin account with a console started without
> >> admin priviledges. So it's always a bit hard to figure out if there's a
> regression.
> >
> > Looks like it's an environment issue. Many tests will likely to fail with a non
> English locale, as error messages from the CRT come localized. Still worth to
> check, could you please pastebin the diffs?
>
> Indeed, I can remember at least some tests failing due to locale issues.
> IIRC, I have fixed some (all?) of them quite long ago.
>
> I'll investigate closer on this issue.
>
> > In any case, checking travis should be a common habit :)
>
> Yes, you're right. It would be nice, though, if an automatic mail could be sent to
> the committer in case of failure.
>
It sends now a notification on IRC. But sometime that fails, too :)
Regards
Anatol
> Cheers!
>
> > Thanks
> >
> > Anatol
> >
> >> I'll be more careful, though, next time.
> >>
> >> Cheers!
> >>
> >>> Cheers
> >>>
> >>> On 30/08/2016 02:05, Christoph Michael Becker wrote:
> >>>> Commit: 207dab585aadd9f320976a0895734d7d9ca7b862
> >>>> Author: Christoph M. Becker <cmbecker69@gmx.de> Tue, 30 Aug
> >>>> 2016 01:53:27 +0200
> >>>> Parents: c7b4cd1a51736b15d095b6b540f17076a0a8433d
> >>>> Branches: PHP-5.6 PHP-7.0 PHP-7.1 master
> >>>>
> >>>> Link:
> >>>>
> >>>> http://git.php.net/?p=php-src.git;a=commitdiff;h=207dab585aadd9f320
> >>>> 97
> >>>> 6a0895734d7d9ca7b862
> >>>>
> >>>>
> >>>> Log:
> >>>> Fix #71882: Negative ftruncate() on php://memory exhausts memory
> >>>>
> >>>> We must not pass negative sizes to a size_t parameter.
> >>>>
> >>>> Bugs:
> >>>> https://bugs.php.net/71882
> >>>>
> >>>> Changed paths:
> >>>> M NEWS
> >>>> M ext/standard/file.c
> >>>> A ext/standard/tests/file/bug71882.phpt
> >>>>
> >>>>
> >>>> Diff:
> >>>> diff --git a/NEWS b/NEWS
> >>>> index bc3d279..a05fa84 100644
> >>>> --- a/NEWS
> >>>> +++ b/NEWS
> >>>> @@ -48,6 +48,8 @@
> >>>> PHP
> >>>> NEWS
> >>>> . Fixed bug #72278 (getimagesize returning FALSE on valid jpg). (cmb)
> >>>> . Fixed bug #65550 (get_browser() incorrectly parses entries
> >>>> with "+" sign).
> >>>> (cmb)
> >>>> + . Fixed bug #71882 (Negative ftruncate() on php://memory
> >>>> + exhausts
> >>>> memory).
> >>>> + (cmb)
> >>>>
> >>>> - XML:
> >>>> . Fixed bug #72085 (SEGV on unknown address zif_xml_parse).
> >>>> (cmb) diff --git a/ext/standard/file.c b/ext/standard/file.c index
> >>>> d8471ff..d346ed9 100644
> >>>> --- a/ext/standard/file.c
> >>>> +++ b/ext/standard/file.c
> >>>> @@ -1512,6 +1512,11 @@ PHP_NAMED_FUNCTION(php_if_ftruncate)
> >>>> RETURN_FALSE;
> >>>> }
> >>>>
> >>>> + if (size < 0) {
> >>>> + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Negative size
> >>>> + is
> >>>> not supported");
> >>>> + RETURN_FALSE;
> >>>> + }
> >>>> +
> >>>> PHP_STREAM_TO_ZVAL(stream, &fp);
> >>>>
> >>>> if (!php_stream_truncate_supported(stream)) { diff --git
> >>>> a/ext/standard/tests/file/bug71882.phpt
> >>>> b/ext/standard/tests/file/bug71882.phpt
> >>>> new file mode 100644
> >>>> index 0000000..ae0137b
> >>>> --- /dev/null
> >>>> +++ b/ext/standard/tests/file/bug71882.phpt
> >>>> @@ -0,0 +1,11 @@
> >>>> +--TEST--
> >>>> +Bug #71882 (Negative ftruncate() on php://memory exhausts memory)
> >>>> +--FILE--
> >>>> +<?php
> >>>> +$fd = fopen("php://memory", "w+"); ftruncate($fd, -1);
> >>>> ?> ==DONE==
> >>>> +--EXPECTF--
> >>>> +Warning: ftruncate(): Negative size is not supported in
> >>>> %s%ebug71882.php on line %d
> >>>> +==DONE==
> >>>>
> >>>>
> >>>
> >>>
> >>
> >> --
> >> PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit:
> >> http://www.php.net/unsub.php
> >
> >
> >
>
> --
> PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit:
> http://www.php.net/unsub.php