This is a list of open issues that need to be resolved for FreeBSD
5.1. If you have any updates for this list, please e-mail
re@FreeBSD.org.
| Issue | Status | Responsible | Description |
|---|
| nullfs deadlocks |
-- |
-- |
&a.kris; reports deadlocks involving the use of nullfs in the bento
environment: buildworld -j4 with src and obj mounted via nullfs; the gcc
processes eventually deadlocked in the ufs state. DDB traceback
showed two different codepaths. I've just repeated this, so the bug
still exists. |
| SMP double fault |
-- |
-- |
&a.kris; reports that Bento (SMP machine) is regularly crashing
with a double fault, or just locking up solidly (cannot break to DDB
on the console). This may be hardware-related, though, because
bento also panicked recently with an NMI. I'm currently running it
UP to see if the problems persist. &a.jake; also suggested increasing
the size of the kernel stack in case that is the cause of the double
faults. |
| Spurious alpha panics |
-- |
-- |
&a.kris; reports that he has seen several unique panics on the
alpha machines, but since gdb -k doesn't work on alpha I haven't
been able to get more than a DDB traceback. No-one has looked
into them. |
| gdb -k doesn't work on alpha |
-- |
&a.mp; |
gdb -k doesn't work on alpha |
| TIMEWAIT panic for network labeled MAC policy modules |
In progress |
&a.rwatson; |
Due to TCP state efficiency improvements in the TIMEWAIT state,
it's now possible for the PCB socket pointer to be NULL in
situations it previously wasn't. As a result, TCP responses
generated in the TIMEWAIT state may lack properly initialized
MAC labels, resulting in panics for network labeled MAC policies.
A fix is current in testing. |
| ipfw/ipfw2 alignment issues on alpha/sparc64 |
In progress |
&a.luigi; |
There are reports of alignment problems with ipfw and/or ipfw2
on 64-bit platforms (specifically alpha and sparc64). |
| ia64 stability |
-- |
-- |
&a.kris; reports high instability of 5.0-CURRENT on ia64
machines, such as the pluto* machines. These problems need to be
fixed in order to get a successful package build. |
| gcc sparc64 problems |
-- |
-- |
&a.kris; reports that there is a gcc bug on sparc64 that
prevents the build of XFree86-4-libraries. &a.kan; is working on
a gcc 3.3 patchset for test. |
| ia64 sio support |
In progress |
&a.marcel;, &a.imp; |
ia64 serial console support is not currently functional,
and requires a reworking of the sio support to improve
platform independence and bus handling. |
| MAC Framework devfs path fixes |
In progress |
&a.rwatson; |
&a.green; has submitted patches to improve the consistency
of the pathnames passed into the MAC Framework devfs labeling
entry points. These patches need to be thoroughly reviewed
and tested, then merged. |
| MAC Framework locking performance improvements |
In progress |
&a.rwatson; |
MAC Framework synchronization currently costs two mutex lock
and unlock operations for each entry point, even if no
dynamically loadable/unloadable policies are present. Patches
are currently being tested that halve this locking cost when
dynamic policies are not loaded. |
| Issue | Status | Responsible | Description |
|---|
| a.out toolchain port |
-- |
-- |
A.out support in the base system was reduced; the removed build
chain components should be made into a port. |
| Race conditions in truss |
Errata candidate |
&a.robert; |
Truss appears to contain a race condition during the start-up of
debugging, which can result in truss failing to attach to the process
before it exits. The symptom is that truss reports that it cannot
open the procfs node supporting the process being debugged. A bug
also appears to exist where in truss will hang if execve() returns
ENOENT. A further race appears to exist in which truss will return
"PIOCWAIT: Input/output error" occasionally on startup. The fix
for this sufficiently changes process execution handling that we
will defer the fix to post-5.0 and consider this errata. |
| KSE support for sparc64, alpha, ia64 |
-- |
&a.jake;, --, -- |
Currently, the MD elements of KSE are present only for the i386
platform, limiting use of KSE to the i386 platform. It is highly
desirable to make KSE available on non-i386 platforms for
5.1-RELEASE so that KSE can see more broad exposure, and the
performance benefits of KSE can be visible to users of the
64-bit FreeBSD architectures. |
-
- | 16-byte CDB addressing for CAM |
- In progress |
- &a.ken; |
- CAM currently supports only 12-byte CDBs, placing a 2TB bound
- on addressable disk storage using CAM. With support for large
- file systems with UFS2, and large partitions using GPT, CAM
- now needs to be updated to support these devices. |
-
-
| Issue | Status | Responsible | Description |
|---|
+
+ | 16-byte CDB addressing for CAM |
+ -- |
+ -- |
+ 16 byte CDB support was recently added to CAM. This allows
+ access to volumes larger than 2TB (2^32 blocks), though few
+ devices have been found that support this. The new code can be
+ tested with the following command, though doing this on a device
+ that does not support 16 byte CDB's will likely result in major
+ problems: sysctl -w kern.cam.da.0.minimum_cmd_size=16 |
+
+
| Fresh ACPI-CA import |
-- |
-- |
The 20030228 vendor sources have been imported. Further testing is
appreciated. |
| iir driver fails to probe disks |
-- |
-- |
During the install, the iir controller appears to correctly
probe, but finds no disk devices. The card in question is an
Intel srcu32 RAID controller with 32Mb RAM, Firmware revision
2.33.01. Intel developers have released an update driver, which
has recently been adapted and imported into 5-CURRENT, and must
be tested. |
| PAE support for i386 |
-- |
-- |
PAE support allows the use of up to 64GB of RAM on Pentium Pro and above
systems. Virtual addresses are still constrained to 32-bits. |
-
- | lazy context switching for i386 |
- -- |
- -- |
- Lazy Context Switching allows for faster context switches and should
- mitigate the some of the latency introduced with interrupt threads and
- GEOM kernel threads. Enabled with the LAZY_SWITCH option. |
-
-
| if_wi problems on Lucent hardware |
-- |
-- |
The recently upgraded if_wi driver is more tuned to Prism hardware
than to Lucent hardware, resulting in system lockups and poor performance
when using Lucent hardware. These problems are believed to be fixed
but more testing is welcome. |
| No console (term) if smbus is enable in the kernel |
-- |
-- |
There are reports that kernels build with smbus, viapm, and smb
devices may not properly probe and support the system console
following boot. |
| UFS2 as installation, newfs default |
-- |
&a.rwatson; |
For 5.1-RELEASE, the default file system type for newly created
file systems is UFS2 rather than UFS1. newfs(8) and sysinstall(8)
have been updated to use this new default. Testing to make sure all
goes well after the change (committed on April 20, 2003) is
vital. |
| wait4 "impossible" mutex panic |
-- |
&a.tjr; |
&a.kris; reports mutex panics in wait4 in the bento environment:
"Impossible" mutex panic in wait4() (reported to -current a while
ago). Believed to be related to (and fixed along with) uidinfo
"impossible" panic below. |
| uidinfo "impossible" panic |
-- |
&a.tjr; |
&a.kris; reports that there appears to be a resource leak of
uidinfo reference counts that is causing the gohan machines to panic
after about 18 days of load. &a.tjr; has committed a fix that is
believed to solve the problem. |
| NSSwitch support |
-- |
&a.nectar; |
Support for pluggable directory services using NSS, including
adaptations of current directory services (local databases, NIS),
and support for new services (LDAP, Active Directory, etc).
This change has been committed, and requires broader testing. |
&footer;