negative values for sizes >2gb

[Jeff Johnson]

tracker

[Serge (serge-redhat-bugs)]

Steps to Reproduce:
1. yum localinstall 3gb rpm

Actual results:
yum prints Size -1075798307.0, Total size: -1075605535

Expected results:

positive values in 3gb range

[seth (seth-redhat-bugs)]

oooo - that looks like fun.

Can you cut and paste the whole output you're seeing - I want to make sure I have the context right.

also could you run:

rpm -qpi 3gb.rpm and paste that output, too.

[Serge (serge-redhat-bugs)]

[root@localhost i686]# yum localinstall quake4-1.4.2-3pal.i686.rpm
Loaded plugins: aliases, auto-update-debuginfo, basearchonly, changelog, downloadonly, fastestmirror, filter-data, keys, list-data, merge-conf, post-transaction-actions, presto,
              : priorities, protect-packages, protectbase, refresh-packagekit, refresh-updatesd, remove-with-leaves, rpm-warm-cache, security, show-leaves, tmprepo, tsflags, upgrade-
              : helper, verify, versionlock
Setting up Local Package Process
Examining quake4-1.4.2-3pal.i686.rpm: quake4-1.4.2-3pal.i686
Marking quake4-1.4.2-3pal.i686.rpm to be installed
Found 128 installed debuginfo package(s)
0 packages excluded due to repository protections
Loading mirror speeds from cached hostfile
 * rawhide: mirror.karneval.cz
 * rawhide-debuginfo: mirror.karneval.cz
 * rpmfusion-free-rawhide: mirror.karneval.cz
 * rpmfusion-free-rawhide-debuginfo: ftp-stud.hs-esslingen.de
 * rpmfusion-nonfree-rawhide: mirror.karneval.cz
 * rpmfusion-nonfree-rawhide-debuginfo: mirror.karneval.cz
0 packages excluded due to repository protections
Resolving Dependencies
--> Running transaction check
---> Package quake4.i686 0:1.4.2-3pal set to be updated
--> Processing Dependency: libSDL-1.2.so.0 for package: quake4-1.4.2-3pal.i686
--> Running transaction check
---> Package SDL.i686 0:1.2.13-10.fc12 set to be updated
--> Finished Dependency Resolution

Dependencies Resolved

===========================================================================================================================================================================================
 Package Арх. Версия Repository Size
===========================================================================================================================================================================================
Installing:
 quake4 i686 1.4.2-3pal /quake4-1.4.2-3pal.i686 -1075798307.0
Installing for dependencies:
 SDL i686 1.2.13-10.fc12 rawhide 188 k

Transaction Summary
===========================================================================================================================================================================================
Install 2 Package(s)
Upgrade 0 Package(s)

Total size: -1075605535
Total download size: 188 k
Выполнить [y/N]: y

[Serge (serge-redhat-bugs)]

[root@localhost i686]# rpm -qpi quake4-1.4.2-3pal.i686.rpm
Name : quake4 Relocations: (not relocatable)
Version : 1.4.2 Vendor: (none)
Release : 3pal Build Date: Вск 09 Авг 2009 16:44:15
Install Date: (not installed) Build Host: vpn-107-53.tenet.odessa.ua
Group : Amusements/Games Source RPM: quake4-1.4.2-3pal.src.rpm
Size : 3219168989 License: Proprietary
Signature : (none)
URL : http://www.idsoftware.com/
Summary : Quake 4
Description :
Quake 4

[Panu (panu-redhat-bugs)]

(In reply to comment #1)
> oooo - that looks like fun.

This is indeed going to be fun on several fronts... some of it in createrepo and yum which seem to assume int() for size in various places, and some in rpm-python which doesn't deal with signed/unsigned issues either.

And then there's the land beyond 4GB sizes which isn't dealt with at all on the python side of things currently, but that's another (but related) story.

[seth (seth-redhat-bugs)]

well changing it in yum should fix createrepo but in this case it is just in yum/rpm-python afaict b/c this pkg is being localinstalled.

I'm building myself a 3gb rpm to see what all falls apart.

[seth (seth-redhat-bugs)]

definitely in rpm-python for this part of this one. I'll see if I can come up with a sensible answer.

[Panu (panu-redhat-bugs)]

The issue (well, one of them) is that python integers are always signed, whereas rpm itself mostly deals with unsigned (32bit) integers. Hardly an issue on 64bit systems but on 32bit systems you get results like this.

To fix it, rpm-python needs to start returning python longs instead of ints, which in turn will break anything assuming integers or have some less dramatic side-effects like unexpected L's in output etc.

[Serge (serge-redhat-bugs)]

my system is 64bit, im just installing 32bit package

[seth (seth-redhat-bugs)]

This patch seems to work
--- a/python/header-py.c~ 2009-06-23 07:40:59.000000000 -0400
+++ b/python/header-py.c 2009-10-27 12:07:08.000000000 -0400
@@ -390,13 +390,13 @@
    if (count != 1 || forceArray) {
        metao = PyList_New(0);
        for (i = 0; i < count; i++) {
- o = PyInt_FromLong(((int *) data)[i]);
+ o = PyLong_FromUnsignedLong(((int *) data)[i]);
        PyList_Append(metao, o);
        Py_DECREF(o);
        }
        o = metao;
    } else {
- o = PyInt_FromLong(*((int *) data));
+ o = PyLong_FromUnsignedLong(*((int *) data));
    }
    break;

I've tested installs and queries and createrepo and all of them seem to return the right info.

[Bug (bug-redhat-bugs)]

This bug appears to have been reported against 'rawhide' during the Fedora 12 development cycle.
Changing version to '12'.

More information and reason for this action is here:
http://fedoraproject.org/wiki/BugZappers/HouseKeeping

[Fedora (fedora-redhat-bugs)]

rpm-4.7.2-1.fc12 has been submitted as an update for Fedora 12.
http://admin.fedoraproject.org/updates/rpm-4.7.2-1.fc12

[Fedora (fedora-redhat-bugs)]

rpm-4.7.2-1.fc12 has been pushed to the Fedora 12 testing repository. If problems still persist, please make note of it in this bug report.
 If you want to test the update, you can install it with
 su -c 'yum --enablerepo=updates-testing update rpm'. You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F12/FEDORA-2009-12988

[Panu (panu-redhat-bugs)]

Closing this, the update got long since published but bodhi seems to have hickuped or something...