Bug #922
SUNWdtrt needs some love
Start date:
2011-04-18
Due date:
% Done:
100%
Estimated time:
Difficulty:
Tags:
Gerrit CR:
Description
There are various failures in SUNWdtrt due to the tests not having kept up with the system. Changes to binaries that it expects certain predictable behaviour from, changes to system calls, and similar things have induced a variety of bogus failures that make testing DTrace changes harder than it should be.
Updated by Rich Lowe over 9 years ago
- Category set to cmd - userland programs
- Status changed from New to In Progress
The issues are
- umod/usym/ufunc tests all assume an infinite loop incrementing a counter will likely cause the program counter to be on an instruction within 'ksh' when a profile probe fires. ksh93 makes this unlikely or impossible
- A test of globbing was not updated when the readlinkat system call appeared
- the violentdeath test inadvertently would kill the wrong process, leaving DTrace running and grinding the system
- tst.include.ksh was not updated to account for several troublesome headers
rds.h: makes use of
__attribute__(packed)
smbios_impl.h: Uses thread-local storage by way of string.hwhich should be added to the blacklist, and:
fs_reparse.h: Uses 'string' as an argument name, and trips the DTrace parsers use of the same.
Where I plan to correct the argument name (such that the structures and such within are more easily available to DTrace)
- tst.spin.sh runs dirname with an argument likely to be the empty string, which only works with /bin/dirname, so use it explicitly.
Updated by Rich Lowe over 9 years ago
- Status changed from In Progress to Resolved
- % Done changed from 0 to 100
Resolved in r13350 commit:a4d1acb7a7f5