As part of preparing to leave, I have filed a few more bugs.
There are not many, because most of our porting problems are
processor-specific or platform-specific, and do not uncover
defects in common code, which already has a pretty reasonable
track record of working on various Sparc models, i386, AMD64,
and even the old Solaris/PPC 2.6.
As for our PowerPC problems, we have not used a bug tracking
system for our internal purposes, with Sun Labs.
I have collected the common code bugs that have been filed,
so far. Some are outright bugs; others are RFEs (Request
For Enhancement), where I do not argue that Solaris common
code is defective, I am just hoping to convince others to
add some functionality or improve performance or robustness.
It would make Solaris better, overall, in my opinion, and
it would make it easier to keep Polaris and Solaris in sync.
Here is a summary of the Polaris-related bugs.
Bug id Synopsis
------- --------------------------------------------------
6425059 Typographical error in ccompile.h breaks builds using GNU C
6506506 spa_history.c: LE_64(reclen) needs to be cast to uint64_t
in case of 32-bit big-endian kernel
6425061 thread.h t_nomigrate, naked char type is not portable,
should be signed
6634371 Solaris ON is broken w.r.t. 64-bit operations on 32-bit processors
6663275 mod_containing_pc() should survive, even if no modules have
been loaded
6663297 vmem_contains() can terminate search earlier
6663313 Proposed new function: vmem_contains_alloc()
6663335 Smaller, faster __four_digits_quick()
The last 4 were filed today, so you might not see them show up on
opensolaris.org, yet. I don't know how much delay to expect.
-- Guy Shaw
