Bug#492377: mksh: strcasestr implicitly converted to pointer
On Mon, Jul 28, 2008 at 09:15:01PM +, Thorsten Glaser wrote: Hi! implicit_pattern = re.compile(([^:]*):(\d+): warning: implicit declaration ??? I think that a simple addition to the already-existing sed command, like s/^[^:]*:[0-9]*:/config test;/ (which I did in mksh-35.2-2, pending to be uploaded by my sponsor??) should avoid your script to be triggered. Well, I'm not too worried about my copy getting triggered - I can currently remember the pkgs w/ false positives. But, this might be a good thing to add to the buildd's copy (maintainer cc'd). Note that the sed is only run on configure-time $CC calls anyway, not on calls to $CC used to really build mksh with the compiled results, so that real bugs in the source were to still be caught, if they then existed. ??? I just realised I have a freeze exception on mksh, even if I don???t know why, so let???s use it??? (I learned about the freeze 2 days ago on the media???) looks like you just did, thanks! bye, //mirabilos -- dann frazier -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#492377: mksh: strcasestr implicitly converted to pointer
Hi! implicit_pattern = re.compile(([^:]*):(\d+): warning: implicit declaration … I think that a simple addition to the already-existing sed command, like s/^[^:]*:[0-9]*:/config test;/ (which I did in mksh-35.2-2, pending to be uploaded by my sponsor¹) should avoid your script to be triggered. Note that the sed is only run on configure-time $CC calls anyway, not on calls to $CC used to really build mksh with the compiled results, so that real bugs in the source were to still be caught, if they then existed. ① I just realised I have a freeze exception on mksh, even if I don’t know why, so let’s use it… (I learned about the freeze 2 days ago on the media…) bye, //mirabilos -- Sometimes they [people] care too much: pretty printers [and syntax highligh- ting, d.A.] mechanically produce pretty output that accentuates irrelevant detail in the program, which is as sensible as putting all the prepositions in English text in bold font. -- Rob Pike in Notes on Programming in C -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#492377: mksh: strcasestr implicitly converted to pointer
Package: mksh Version: 35.2-1 Severity: serious Usertags: implicit-pointer-conversion Our automated buildd log filter[1] detected a problem that is likely to cause your package to segfault on architectures where the size of a pointer is greater than the size of an integer, such as ia64 and amd64. Function `strcasestr' implicitly converted to pointer at ] scn.c:4 This is often due to a missing function prototype definition. For more information, see [2]. Though it is guaranteed that this codepath will cause a segfault on certain architectures, it is not guaranteed that this codepath would ever be executed (e.g., if the returned pointer is never dereferenced). However, this bug does prevent the ia64 buildd from successfully building this package, resulting in a practical FTBFS issue and warranting the serious severity. [1] http://people.debian.org/~dannf/check-implicit-pointer-functions [2] http://wiki.debian.org/ImplicitPointerConversions -- dann frazier -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#492377: mksh: strcasestr implicitly converted to pointer
tags 492377 + help reassign 492377 qa.debian.org thanks dann frazier dixit: Function `strcasestr' implicitly converted to pointer at ] scn.c:4 I suggest you read your eMail INBOX, as I already wrote to you in Message-ID: [EMAIL PROTECTED] that this is a FALSE POSITIVE and that I request help for how to deal with this bug in your automated checker. Excerpt from that eMail, for the archives: │http://buildd.debian.org/fetch.cgi?pkg=mksharch=ia64ver=35.2-1stamp=1216828593file=logas=raw │ │The mksh build process involves a configuration and a build step, similar │to what GNU autoconf does. (In fact, it does this twice, once for the │normal glibc binary, once for a small dietlibc binary.) However, in con- │trast to GNU autoconf, the configuration step is verbose. │ │The autoconf check for strcasestr() is expected to return no with dietlibc, │because the function is not defined. Since it is not declared in any sy- │stem header file (of dietlibc), we have no prototype, so the compile test │also triggers a false positive on the warning: │Diagnostic output from the autoconf checks is prepended by “] ” (and always │occurs on the file scn.c, akin to conftest.c in GNU autoconf). │ │What do you suggest to remove this false positive while keeping the confi- │guration process as warning-free as possible? As for a possible solution, I don’t think that adding an exemption of mksh to the checker is worth the effort, but I can’t think of a modifi- cation to the test that would NOT trigger the warning while staying warning-free in all other cases (as the test is EXPECTED to fail on Debian with Fefe’s dietlibc). Maybe I could play some sed magic to fool the regex your scanner uses, so that it doesn’t recognise warnings emitted during the configuration process? This would then be a Debian- specific kludge and NOT go in upstream. If you have any BETTER solutions, please let me know. bye, //mirabilos -- Sometimes they [people] care too much: pretty printers [and syntax highligh- ting, d.A.] mechanically produce pretty output that accentuates irrelevant detail in the program, which is as sensible as putting all the prepositions in English text in bold font. -- Rob Pike in Notes on Programming in C -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#492377: mksh: strcasestr implicitly converted to pointer
reassign 492377 mksh thanks The seriousness of this bug is actually caused by the fact that we will be unable to do ia64 updates of mksh because the ia64 buildd will reject them. This doesn't mean it needs to be fixed in mksh of course, but mksh cannot release until it is fixed somehow, so the bug should remain here. On Fri, Jul 25, 2008 at 06:57:43PM +, Thorsten Glaser wrote: As for a possible solution, I don???t think that adding an exemption of mksh to the checker is worth the effort, but I can???t think of a modifi- cation to the test that would NOT trigger the warning while staying warning-free in all other cases (as the test is EXPECTED to fail on Debian with Fefe???s dietlibc). Maybe I could play some sed magic to fool the regex your scanner uses, so that it doesn???t recognise warnings emitted during the configuration process? This would then be a Debian- specific kludge and NOT go in upstream. If you have any BETTER solutions, please let me know. I don't have any good solutions in mind, and I also don't like the idea of you having to change somethinto workaround this filter. Maybe the buildd maintainer (CC'd) could install a whitelist file or something? -- dann frazier -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#492377: mksh: strcasestr implicitly converted to pointer
dann frazier dixit: but mksh cannot release until it is fixed somehow Hm okay. I don't have any good solutions in mind Good, then I might try some workaround against the regex matcher. Can a non-DD get access to an IA64 test machine, or can I send a beta package for testing and NOT uploading to someone, once I hacked that potential workaround? TIA, //mirabilos -- Sometimes they [people] care too much: pretty printers [and syntax highligh- ting, d.A.] mechanically produce pretty output that accentuates irrelevant detail in the program, which is as sensible as putting all the prepositions in English text in bold font. -- Rob Pike in Notes on Programming in C -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#492377: mksh: strcasestr implicitly converted to pointer
Thorsten Glaser [EMAIL PROTECTED] (25/07/2008): dann frazier dixit: but mksh cannot release until it is fixed somehow Hm okay. Putting him back in the loop, submitters don't receive mails to the bug they opened (thanks Don for that…). I don't have any good solutions in mind Good, then I might try some workaround against the regex matcher. Can a non-DD get access to an IA64 test machine, or can I send a beta package for testing and NOT uploading to someone, once I hacked that potential workaround? Check http://db.debian.org/machines.cgi, it looks like merulo would be accessible by anyone? Mraw, KiBi. signature.asc Description: Digital signature
