[pacman-dev] Segmentation fault
Hi! Finally I could reproduce FS#8350. Some background: -this is probably not important, but my "union" sync repo is sitting on my pendrive [in the suggestive file:///p/1 path ;-], so my syncdb is always out of sync [pendrive's size-limitation] -tk is in my sync repo [/p/1], cdparanoia is not [neither of the packages is installed, both can be found in my syncdb] S_good.txt: # pacman -S cdparanoia tk --debug S_fail.txt # pacman -S cdparanoia tk --debug Hm. What's the difference?: In the first case tk is in my /var/cache, so pacman tries to download only cdparanoia. In the second case my cache is empty (after -Scc), so pacman tries to download both (however, pacman could download tk to my cache before segfault) Bye, ngaba ---------------------------------------------------- SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu This mail sent through IMP: http://horde.org/imp/
On Nov 15, 2007 4:10 AM, Nagy Gabor <ngaba@bibl.u-szeged.hu> wrote:
Hm. What's the difference?: In the first case tk is in my /var/cache, so pacman tries to download only cdparanoia. In the second case my cache is empty (after -Scc), so pacman tries to download both (however, pacman could download tk to my cache before segfault)
Hmm, right off the bat I noticed this: error: failed retrieving file 'cdparanoia-9.8-4.pkg.tar.gz' from : No such file or directory That is apparently a blank server... we should take this into account, but to verify if this is the bug or not, can you remove that empty server from your config and see if it changes anything?
On Nov 15, 2007 10:55 AM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Nov 15, 2007 4:10 AM, Nagy Gabor <ngaba@bibl.u-szeged.hu> wrote:
Hm. What's the difference?: In the first case tk is in my /var/cache, so pacman tries to download only cdparanoia. In the second case my cache is empty (after -Scc), so pacman tries to download both (however, pacman could download tk to my cache before segfault)
Hmm, right off the bat I noticed this:
error: failed retrieving file 'cdparanoia-9.8-4.pkg.tar.gz' from : No such file or directory
That is apparently a blank server... we should take this into account, but to verify if this is the bug or not, can you remove that empty server from your config and see if it changes anything?
For that matter, can you zip up your pacman.conf and any mirrorlist files that are included from pacman.conf? Maybe we can follow the parsing that way. -Dan
Idézés Dan McGee <dpmcgee@gmail.com>:
On Nov 15, 2007 4:10 AM, Nagy Gabor <ngaba@bibl.u-szeged.hu> wrote:
Hm. What's the difference?: In the first case tk is in my /var/cache, so pacman tries to download only cdparanoia. In the second case my cache is empty (after -Scc), so pacman
On Nov 15, 2007 10:55 AM, Aaron Griffin <aaronmgriffin@gmail.com> wrote: tries
to download both (however, pacman could download tk to my cache before segfault)
Hmm, right off the bat I noticed this:
error: failed retrieving file 'cdparanoia-9.8-4.pkg.tar.gz' from : No such file or directory
That is apparently a blank server... we should take this into account, but to verify if this is the bug or not, can you remove that empty server from your config and see if it changes anything?
For that matter, can you zip up your pacman.conf and any mirrorlist files that are included from pacman.conf? Maybe we can follow the parsing that way.
Attached. Bye ---------------------------------------------------- SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu This mail sent through IMP: http://horde.org/imp/
On Nov 15, 2007 10:55 AM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Nov 15, 2007 4:10 AM, Nagy Gabor <ngaba@bibl.u-szeged.hu> wrote:
Hm. What's the difference?: In the first case tk is in my /var/cache, so pacman tries to download only cdparanoia. In the second case my cache is empty (after -Scc), so pacman tries to download both (however, pacman could download tk to my cache before segfault)
Hmm, right off the bat I noticed this:
error: failed retrieving file 'cdparanoia-9.8-4.pkg.tar.gz' from : No such file or directory
That is apparently a blank server... we should take this into account, but to verify if this is the bug or not, can you remove that empty server from your config and see if it changes anything?
It is actually not a blank server, but a file:// URL which sets the server part to null or "" (hopefully, otherwise we could have a segfault here). Thus, only the path is set. I'm sure we can find a way to make this message more clear. -Dan
On Nov 15, 2007 1:09 PM, Dan McGee <dpmcgee@gmail.com> wrote:
It is actually not a blank server, but a file:// URL which sets the server part to null or "" (hopefully, otherwise we could have a segfault here). Thus, only the path is set. I'm sure we can find a way to make this message more clear.
I noticed that later too... http://code.phraktured.net/?p=pacman.git;a=commitdiff;h=c10aa2d7c9fd37a3f8f0... Still, the original segfault remains.
On Nov 15, 2007 1:21 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Nov 15, 2007 1:09 PM, Dan McGee <dpmcgee@gmail.com> wrote:
It is actually not a blank server, but a file:// URL which sets the server part to null or "" (hopefully, otherwise we could have a segfault here). Thus, only the path is set. I'm sure we can find a way to make this message more clear.
I noticed that later too...
http://code.phraktured.net/?p=pacman.git;a=commitdiff;h=c10aa2d7c9fd37a3f8f0...
Still, the original segfault remains.
Not to be nitpicky, but that could be a false test case (if the server name did not get set for some reason). Why not use strcmp(SCHEME_FILE, url->scheme)? -Dan
On Nov 15, 2007 2:08 PM, Dan McGee <dpmcgee@gmail.com> wrote:
On Nov 15, 2007 1:21 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Nov 15, 2007 1:09 PM, Dan McGee <dpmcgee@gmail.com> wrote:
It is actually not a blank server, but a file:// URL which sets the server part to null or "" (hopefully, otherwise we could have a segfault here). Thus, only the path is set. I'm sure we can find a way to make this message more clear.
I noticed that later too...
http://code.phraktured.net/?p=pacman.git;a=commitdiff;h=c10aa2d7c9fd37a3f8f0...
Still, the original segfault remains.
Not to be nitpicky, but that could be a false test case (if the server name did not get set for some reason). Why not use strcmp(SCHEME_FILE, url->scheme)?
True and all but if there's no host name, and it's not a file url we shouldn't be at that point in the code Either way, I'll change it
participants (3)
-
Aaron Griffin
-
Dan McGee
-
Nagy Gabor