Re: VFS requirements for carbon/cocoa?
Re: VFS requirements for carbon/cocoa?
- Subject: Re: VFS requirements for carbon/cocoa?
- From: Chaskiel Grundman <email@hidden>
- Date: Wed, 29 Jun 2005 21:10:28 -0400
short answer: problem solved. my VFS_READDIR was broken.
--On Wednesday, June 29, 2005 19:57:04 +0100 Quinn <email@hidden>
wrote:
Your test exercises weird code paths (see below). I'd recommend that
you copy the entire home directory, not just the ~/Library folder, to
your file system.
This makes no difference, FWIW
If ATS is wedged, not much on the system is going to work. What does
a "sample" log of ATS look like?
<x-man-page://1/sample>
946 FSGetCatalogInfoBulk
946 PBGetCatalogInfoBulkSync
946 POSIXMount::getattrsbulk(void*, unsigned long, unsigned long*,
unsigned long, unsigned long, FSAttributeInfo*, unsigned char*)
946 GetIndexedDirectoryItem(VolumeInfo*, unsigned long,
unsigned long, unsigned long, char*, unsigned char*)
946 DCBlock_Create(VolumeInfo const*, unsigned long, DCBlock**)
946 DCBlock_BuildContents(int, DCBlock*)
946 getdirentries_exhaustive(int, unsigned long*, long*)
946 getdirentries_exhaustive(int, unsigned long*, long*)
My VFS_READDIR was setting d_reclen to 0 in the last dirent object in the
list. This presumably did bad things to getdirentries_exhaustive.
Once that was fixed, console logins work much better.
Now onto the next problem: keychain corruption and "impossible"
self-deadlocks in unlink.... (the deadlock is clearly an internal-to-afs
problem, as it involves an internal-to-afs lock)
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden