Bug#748831: RFS: crashme/2.7-1 [ITA]
* Dariusz Dwornikowski dariusz.dwornikow...@cs.put.poznan.pl, 2014-05-30,
13:32:
* Added patch fixing memset transposed arguments
Good catch. Please don't forget to forward the patch upstream.
Laszlo caught this but the patch was backported from 2.8.2 of crashme.
So technically upstream forwarded it to me.
Oh. I read “Forwarded: no” as “it should be forwarded, but it's not been
done yet”. In this case “Forwarded: not-needed” would be more adequate;
or even better, use “Origin: upstream”, and then you don't need
Forwarded at all.
I didn't get answer to my question about mprotect(2):
Shouldn't we run this code also on non-Linux architectures? At least
on kfreebsd-amd64, heap is not executable by default, which is what
this code is trying to work around.
Yes, forgot to clear this. I am not sure I fully undestand. Crashme is
built on all archs but this is probably not what you meant.
Won't this conditional catch for kfreebsd-amd64 (defined(__FreeBSD__)
compiler macro will catch it ?
Nope, __FreeBSD__ is not defined on GNU/kFreeBSD, only on “true” FreeBSD
systems.
Here's useful piece of documentation about porting to kFreeBSD:
http://glibc-bsd.alioth.debian.org/porting/PORTING
(the relvant section is “Add our system name to checks here and there”)
Another problem on kFreeBSD is that I get this warning:
crashme.c:758:17: warning: incompatible implicit declaration of built-in
function 'execlp' [enabled by default]
{status = execlp(cmd,cmd,nb,arg2,nt,arg4,arg5,subprocess_ind,NULL);
^
--
Jakub Wilk
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
Oh. I read “Forwarded: no” as “it should be forwarded, but it's not been
done yet”. In this case “Forwarded: not-needed” would be more adequate; or
even better, use “Origin: upstream”, and then you don't need Forwarded at
all.
Thank you for that.
I didn't get answer to my question about mprotect(2):
Shouldn't we run this code also on non-Linux architectures? At least
on kfreebsd-amd64, heap is not executable by default, which is what
this code is trying to work around.
Yes, forgot to clear this. I am not sure I fully undestand. Crashme is
built on all archs but this is probably not what you meant.
Won't this conditional catch for kfreebsd-amd64 (defined(__FreeBSD__)
compiler macro will catch it ?
Nope, __FreeBSD__ is not defined on GNU/kFreeBSD, only on “true” FreeBSD
systems.
Here's useful piece of documentation about porting to kFreeBSD:
http://glibc-bsd.alioth.debian.org/porting/PORTING
(the relvant section is “Add our system name to checks here and there”)
Another problem on kFreeBSD is that I get this warning:
crashme.c:758:17: warning: incompatible implicit declaration of built-in
function 'execlp' [enabled by default]
{status = execlp(cmd,cmd,nb,arg2,nt,arg4,arg5,subprocess_ind,NULL);
^
It appears that package has been sponsored by Laszlo today, I will
tackle porting problem when preparing 2.8.2 upstream release,
meanwhile I will install kfreebsd vm.
You mentioned earlier that 2.7 should be also uploaded to stable
because of [1]. Could you please tell me, or point out a source of
information, how this should be done ? Should I create a debian-stable
branch and prepare a package release stable there (in changelog it
should be stable instead of unstable). Or should I just use
backporting procedures and upload to backports?
[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=749816
--
Dariusz Dwornikowski,
Institute of Computing Science, Poznań University of Technology
www.cs.put.poznan.pl/ddwornikowski/
room 2.7.2 BTiCW | tel. +48 61 665 29 41
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
If my supposition is correct, then this is a grave bug, and after it's fixed in unstable, it should be also fixed in stable and oldstable. I've filed it as #749816. Thanks. * Paul Wise p...@debian.org, 2014-05-27, 08:43: On Tue, May 27, 2014 at 1:07 AM, Dariusz Dwornikowski wrote: I can fix this when I am given DM rights to crashme. Nothing about stable/oldstable updates requires DM rights, you can do that through your normal sponsors. Hear, hear. Just to be clear: I am absolutely not giving anyone DM permissions after a single upload. :-) Ack. Anyway, let's get back to work: Possible typos in the code and in the documentation: ntrys - ntries NTRYS - NTRIES Typo in the changelog: canonincal - canonical I fixed this by updating fix-spelling.patch. +Description: tool to test kernel stability I think s/to test/for testing/ sounds better. - them. Used to test kernel stability. + them. It is used to test kernel's stability. I don't think this use of 's is correct. As a data point, according to Google, “kernel stability” is 40 times more popular than “kernel's stability”. But I'm not a native speaker of English, so I might be saying nonsense here. If you want professional advice about your description, please ask at debian-l10n-english@ldo. I followed here Don'ssuggestion. * Added patch fixing memset transposed arguments Good catch. Please don't forget to forward the patch upstream. Laszlo caught this but the patch was backported from 2.8.2 of crashme. So technically upstream forwarded it to me. The following changes are not documented: - Upstream-Contact update in d/copyright; - d/crashme.install addition; - dropping d/crashme.lintian-overrides; - debian/watch update; - most of changes to d/rules; - package description updates; - README.debian update. I documented it. It should be s/debian/Debian/, BTW. (Although it's not a big deal, as debhelper conveniently renames the file for you, so you correct correct name in the binary package.) Renamed it so debhelper does not have to do it. Let's give it some rest. I didn't get answer to my question about mprotect(2): Shouldn't we run this code also on non-Linux architectures? At least on kfreebsd-amd64, heap is not executable by default, which is what this code is trying to work around. Yes, forgot to clear this. I am not sure I fully undestand. Crashme is built on all archs but this is probably not what you meant. Won't this conditional catch for kfreebsd-amd64 ( defined(__FreeBSD__) compiler macro will catch it ? -- Dariusz Dwornikowski, Institute of Computing Science, Poznań University of Technology www.cs.put.poznan.pl/ddwornikowski/ room 2.7.2 BTiCW | tel. +48 61 665 29 41 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
If my supposition is correct, then this is a grave bug, and after it's fixed in unstable, it should be also fixed in stable and oldstable. I've filed it as #749816. * Paul Wise p...@debian.org, 2014-05-27, 08:43: On Tue, May 27, 2014 at 1:07 AM, Dariusz Dwornikowski wrote: I can fix this when I am given DM rights to crashme. Nothing about stable/oldstable updates requires DM rights, you can do that through your normal sponsors. Hear, hear. Just to be clear: I am absolutely not giving anyone DM permissions after a single upload. :-) Anyway, let's get back to work: Possible typos in the code and in the documentation: ntrys - ntries NTRYS - NTRIES Typo in the changelog: canonincal - canonical +Description: tool to test kernel stability I think s/to test/for testing/ sounds better. - them. Used to test kernel stability. + them. It is used to test kernel's stability. I don't think this use of 's is correct. As a data point, according to Google, “kernel stability” is 40 times more popular than “kernel's stability”. But I'm not a native speaker of English, so I might be saying nonsense here. If you want professional advice about your description, please ask at debian-l10n-english@ldo. * Added patch fixing memset transposed arguments Good catch. Please don't forget to forward the patch upstream. The following changes are not documented: - Upstream-Contact update in d/copyright; - d/crashme.install addition; - dropping d/crashme.lintian-overrides; - debian/watch update; - most of changes to d/rules; - package description updates; - README.debian update. It should be s/debian/Debian/, BTW. (Although it's not a big deal, as debhelper conveniently renames the file for you, so you correct correct name in the binary package.) I didn't get answer to my question about mprotect(2): Shouldn't we run this code also on non-Linux architectures? At least on kfreebsd-amd64, heap is not executable by default, which is what this code is trying to work around. -- Jakub Wilk -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
On 24.05.14 23:38:38, Jakub Wilk wrote: * Dariusz Dwornikowski dariusz.dwornikow...@cs.put.poznan.pl, 2014-05-21, 13:48: If you listed files to remove in debian/clean, you could avoid the override in debian/rules. Yes, fixed that too. Settled for d/rules. Well, now I can't build the source package: rm -f *.o rm -f pddet rm debian/upstream rm: cannot remove ‘debian/upstream’: No such file or directory debian/rules:13: recipe for target 'override_dh_auto_clean' failed I forgot to push sorry, now it will build. What happened to debian/patch/legacy.patch? Upstream writes: 2.7 4-APR-2014 __APPLE__ port, fix linux 64 bit port. I wonder how badly broken is the package in the archive (2.4). My understanding is that it's completely useless on some architectures, because it doesn't actually stress-test anything; but I might be wrong. As upstream informed me legacy.patch is in 2.7 already, which I try to push into Debian, that is why I have deleted it. -- Dariusz Dwornikowski, Institute of Computing Science, Poznań University of Technology www.cs.put.poznan.pl/ddwornikowski/ room 2.7.2 BTiCW | tel. +48 61 665 29 41 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
* Dariusz Dwornikowski dariusz.dwornikow...@cs.put.poznan.pl, 2014-05-26,
17:19:
What happened to debian/patch/legacy.patch?
As upstream informed me legacy.patch is in 2.7 already, which I try to
push into Debian, that is why I have deleted it.
Then this should be documented in the changelog. But at least this part
- {status = execl(cmd,cmd,nb,arg2,nt,arg4,arg5,subprocess_ind,0);
+ {status = execlp(cmd,cmd,nb,arg2,nt,arg4,arg5,subprocess_ind,NULL);
hasn't been merged. And it is really needed; see bug #37304, which your
version of crashme reintroduces.
Upstream writes:
2.7 4-APR-2014 __APPLE__ port, fix linux 64 bit port.
I wonder how badly broken is the package in the archive (2.4). My
understanding is that it's completely useless on some architectures,
because it doesn't actually stress-test anything; but I might be wrong.
If my supposition is correct, then this is a grave bug, and after it's
fixed in unstable, it should be also fixed in stable and oldstable.
--
Jakub Wilk
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
On 26.05.14 18:25:25, Jakub Wilk wrote:
* Dariusz Dwornikowski dariusz.dwornikow...@cs.put.poznan.pl, 2014-05-26,
17:19:
What happened to debian/patch/legacy.patch?
As upstream informed me legacy.patch is in 2.7 already, which I try to
push into Debian, that is why I have deleted it.
Then this should be documented in the changelog. But at least this part
- {status = execl(cmd,cmd,nb,arg2,nt,arg4,arg5,subprocess_ind,0);
+ {status = execlp(cmd,cmd,nb,arg2,nt,arg4,arg5,subprocess_ind,NULL);
hasn't been merged. And it is really needed; see bug #37304, which your
version of crashme reintroduces.
I have updated the git, added the changelog entry, and also a patch
fixing this.
2.7 4-APR-2014 __APPLE__ port, fix linux 64 bit port.
I wonder how badly broken is the package in the archive (2.4). My
understanding is that it's completely useless on some architectures,
because it doesn't actually stress-test anything; but I might be wrong.
If my supposition is correct, then this is a grave bug, and after it's fixed
in unstable, it should be also fixed in stable and oldstable.
I can fix this when I am given DM rights to crashme.
--
Dariusz Dwornikowski,
Institute of Computing Science, Poznań University of Technology
www.cs.put.poznan.pl/ddwornikowski/
room 2.7.2 BTiCW | tel. +48 61 665 29 41
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
On Tue, May 27, 2014 at 1:07 AM, Dariusz Dwornikowski wrote: On 26.05.14 18:25:25, Jakub Wilk wrote: If my supposition is correct, then this is a grave bug, and after it's fixed in unstable, it should be also fixed in stable and oldstable. I can fix this when I am given DM rights to crashme. Nothing about stable/oldstable updates requires DM rights, you can do that through your normal sponsors. https://www.debian.org/doc/manuals/developers-reference/pkgs.html#upload-stable -- bye, pabs http://wiki.debian.org/PaulWise -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
* Dariusz Dwornikowski dariusz.dwornikow...@cs.put.poznan.pl, 2014-05-21, 13:48: If you listed files to remove in debian/clean, you could avoid the override in debian/rules. Yes, fixed that too. Settled for d/rules. Well, now I can't build the source package: rm -f *.o rm -f pddet rm debian/upstream rm: cannot remove ‘debian/upstream’: No such file or directory debian/rules:13: recipe for target 'override_dh_auto_clean' failed What happened to debian/patch/legacy.patch? Upstream writes: 2.7 4-APR-2014 __APPLE__ port, fix linux 64 bit port. I wonder how badly broken is the package in the archive (2.4). My understanding is that it's completely useless on some architectures, because it doesn't actually stress-test anything; but I might be wrong. The relevant fix appears to be: #if defined(__APPLE__) || defined(__FreeBSD__) || defined(linux) /* if we don't do this on the 64-bit architectures then all we get out of our badboy() calls is a nice safe bus error signal. */ int pagesize; pagesize = getpagesize(); if (mprotect((void *)long)data)/pagesize)*pagesize), ((n/pagesize)+1)*pagesize, PROT_READ|PROT_WRITE|PROT_EXEC)) perror(mprotect); #endif Shouldn't we run this code also on non-Linux architectures? At least on kfreebsd-amd64, heap is not executable by default, which is what this code is trying to work around. -- Jakub Wilk -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
Package: sponsorship-requests Severity: normal Dear mentors, I am looking for a sponsor for my package crashme * Package name: crashme Version : 2.7-1 Upstream Author : George Carrette gjcarre...@aol.com * URL : http://people.delphiforums.com/gjc/crashme.html * License : crashme Section : devel It builds those binary packages: crashme- Stress tests operating system stability To access further information about this package, please visit the following URL: http://mentors.debian.net/package/crashme Alternatively, one can download the package with dget using this command: dget -x http://mentors.debian.net/debian/pool/main/c/crashme/crashme_2.7-1.dsc More information about *crashme* can be obtained from http://people.delphiforums.com/gjc/crashme.html. Changes since the last upload: * Imported Upstream version 2.7 * Copyright changed to DEP-5 * New maintainer (Closes: #739083) * Hardening added * Spelling patch refreshed * VCS fields in canonincal format * Bump standards to 3.9.5 * Created manpage for pddet binary * d/rules updated to clean cleanly Regards, -- Dariusz Dwornikowski, Institute of Computing Science, Poznań University of Technology www.cs.put.poznan.pl/ddwornikowski/ room 2.7.2 BTiCW | tel. +48 61 665 29 41 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
Control: owner -1 ! * Dariusz Dwornikowski dariusz.dwornikow...@cs.put.poznan.pl, 2014-05-21, 08:07: dget -x http://mentors.debian.net/debian/pool/main/c/crashme/crashme_2.7-1.dsc [...] * Copyright changed to DEP-5 This is not an accurate description of changes to the copyright file. The original copyright file was already in the DEP-5 format. * New maintainer (Closes: #739083) Did you get any reply from George? * Hardening added That's not an accurate description of the change you did either. (Although enabled hardening might be a side effect of this change.) Why is override_dh_auto_build commented in debian/rules? If this code is not supposed to be run, then remove it. Why do you set CFLAGS in debian/rules? * Spelling patch refreshed Please forward the patch upstream. There's more typos that you might want to fix: $ codespell --skip '*.patch' ./crashme.txt:26: seperate == separate ./crashme.txt:77: enviroment == environment ./crashme.html:147: exersize == exercise ./crashme.html:272: seperate == separate * Bump standards to 3.9.5 Did it require any changes to packaging? * Created manpage for pddet binary I'm sorry to say that, but this manpage is not helpful. I read it twice, and still have no idea what is this program supposed to do. * d/rules updated to clean cleanly If you listed files to remove in debian/clean, you could avoid the override in debian/rules. -- Jakub Wilk -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#748831: RFS: crashme/2.7-1 [ITA]
* Copyright changed to DEP-5 This is not an accurate description of changes to the copyright file. The original copyright file was already in the DEP-5 format. Fixed that. * New maintainer (Closes: #739083) Did you get any reply from George? Yes, we talked that is why I took over the package. We have been working on crashme (he is upstream) past few weeks. * Hardening added That's not an accurate description of the change you did either. (Although enabled hardening might be a side effect of this change.) Why is override_dh_auto_build commented in debian/rules? If this code is not supposed to be run, then remove it. Why do you set CFLAGS in debian/rules? Fixed. Removed. * Spelling patch refreshed Please forward the patch upstream. There's more typos that you might want to fix: $ codespell --skip '*.patch' ./crashme.txt:26: seperate == separate ./crashme.txt:77: enviroment == environment ./crashme.html:147: exersize == exercise ./crashme.html:272: seperate == separate Did that and forwarded to codeplex issue tracker. * Bump standards to 3.9.5 Did it require any changes to packaging? Nope, fixed. * Created manpage for pddet binary I'm sorry to say that, but this manpage is not helpful. I read it twice, and still have no idea what is this program supposed to do. I found a better piece of text to put there, by upstream too. * d/rules updated to clean cleanly If you listed files to remove in debian/clean, you could avoid the override in debian/rules. Yes, fixed that too. Settled for d/rules. The package is in VCS: http://anonscm.debian.org/gitweb/?p=collab-maint/crashme.git -- Dariusz Dwornikowski, Institute of Computing Science, Poznań University of Technology www.cs.put.poznan.pl/ddwornikowski/ room 2.7.2 BTiCW | tel. +48 61 665 29 41 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
