Open Hub
Black Duck Open Hub

 Settings |  Report Duplicate

0
I Use This!
Inactive

Commits : Listings

Analyzed about 2 hours ago. based on code collected about 9 hours ago.
Apr 18, 2023 — Apr 18, 2024
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Added asynchronous IO copy file test and changed the Makefile to build it..
lward
More... over 9 years ago
Fixed format in a cprintf, it had a missing percent-sign.
lward
More... almost 14 years ago
Fixed for 32-bit build. The 64-bit getdirentries routines in both the regular source and the file handles interface assumed 64-bit _SYSIO_OFF_T. Fixed, now. They will return (-1, EINVAL) if given a base pointer outside what is supported by the internal offset.
lward
More... over 14 years ago
Added ifsync and ifdatasync. These are asynchronous versions of the standard calls. In fact, the core has been reworked so that the async versions are the core implementation.
lward
More... over 14 years ago
need _getdirentries symbol for catamount build
rklundt
More... over 14 years ago
Depending on whether I_ASSOCIATIONS was defined the PB_SET_ASSOC macro was being built in two pretty different ways. With this change, the majority of the logic is the same and only whether or not to perform the pertinent list manipulations is different.
lward
More... over 14 years ago
There was an ifdef dependent on I_ASSOCIATIIONS. Notice anything funny about that? Fixed the typo and it builds with the to-be-included code now.
lward
More... over 14 years ago
In the submount routine, at the end we were releasing the pnode we just tried to cover. That should have been a PUT. It is now :)
lward
More... over 14 years ago
Split native_stat into two functions. One works as before. The other, native_simple_stat just does a stat and returns the result in the passed buffer, updating no existing records.
lward
More... over 14 years ago
Make lint happy. The char count variable is only used for symlinks but lint can't tell. Just initialize it always.
lward
More... over 14 years ago
In fhi_iowait, the return code could have been derived from an uninitialized variable if the handle was invalid or the operation not yet complete. Fixed.
lward
More... over 14 years ago
Modified to pass lint checks. There was a claim that a field in a record was possibly used before initialization. The reality was that enough of the record was initialized to cause the rest to be superfluous at initialization. Lint couldn't tell, of course, so we just initialize the entire record now.
lward
More... over 14 years ago
fix obsolete reference to lh_first in PB_DEBUG code block
rklundt
More... over 14 years ago
ND_NXMNT prevents traversal of a mount, so, automounts are not helpful in the presence of this flag.
rklundt
More... over 14 years ago
In open, parent was always being put when we were done with it. This should only have been done when the lock was acquired. I.e., it's only needed when ND_WANTPARENT was passed to namei. Fixed.
lward
More... over 14 years ago
The biglock symbol was not being exported unless SYSIO_TRACING was defined. Moved it out of the dependent block and pulled SYSIO_{ENTER,LEAVE} out as well.
lward
More... over 14 years ago
Reformatted to fit style and tab widths in the rest of the source. No functional changes.
lward
More... over 14 years ago
Had left the old pb_prune routine, the recursive list-based implementation, but ifdef'd out. It's just gone, now.
lward
More... over 14 years ago
The children of a directory are now maintained as a tree, instead of as a list.
lward
More... over 14 years ago
The splay routine is now public, called _sysio_splay. Note, it's still a really bad idea to depend on the other tree routines maintaining their trees as a splay-tree.
lward
More... over 14 years ago
The new logic in pathwalk introduced an error; Reported by Jason Cope at ANL. If a direcotry to be traversed did not grant X permission then the error was caught, the done flag set, and nd->nd_pno is left as NULL. The next thing that was done was to check that the parent mount was same as nd->nd_pno mount. Oops, nd->nd_pno was left set to NULL, resulting in the train wreck we know so well as SIGSEGV.
lward
More... over 14 years ago
Modified to include the new fhi tests.
lward
More... over 14 years ago
Modified to use the new fhi test support goodies.
lward
More... over 14 years ago
New support for fhi tests.
lward
More... over 14 years ago
New test.
lward
More... over 14 years ago
When checking permissions do not always return -EPERM;. Instead, return the error that _sysio_permitted gave to us.
lward
More... over 14 years ago
When I moved the permission check to the p-node routine it was incomplete. Now, check permission and return error if there was one, otherwise continue on with the operation.
lward
More... over 14 years ago
The fhi_symlink arguments are now from (a path) to (a diroparg) instead of the other way round, which was wrong.
lward
More... over 14 years ago
Both the getattr and rmdir routines could return early without going through SYSIO_INTERFACE_RETURN. This had the unfortunate side-effect of leaving the biglock in the locked state :( Fixed.
lward
More... over 14 years ago
It's official. I can't do token pasting to save my life.
lward
More... over 14 years ago