diff --git a/en/releases/5.0R/todo.sgml b/en/releases/5.0R/todo.sgml index a2677aa941..dab6f591c3 100644 --- a/en/releases/5.0R/todo.sgml +++ b/en/releases/5.0R/todo.sgml @@ -1,324 +1,307 @@ - + %includes; %developers; ]> &header;

Open Issues

This is a list of open issues that need to be resolved for FreeBSD 5.0. If you have any updates for this list, please e-mail re@FreeBSD.org.

Must Resolve Issues for 5.0-RC3

- - - - - - -
IssueStatusResponsibleDescription
Many reported problems with disklabel -e -- &a.phk;, &a.jhb; Changes in the disk label format have resulted in substantial numbers of bug reports; these must be resolved by RC1. The problem involving a cache coherency issue relating to partition resizes has now been resolved. Problems relating to disklabel -r and the change appear to continue.
Audit of devfs device permissions In progress &a.kris;,&a.rwatson; With the move to devfs, the permissions specified in kernel device creation lines using make_dev() are now used as the default permissions for /dev entries. Some of these entries are more lax than is desired, or than the MAKEDEV defaults in 4.x. These need to be carefully reviewed and fixed.
mly driver hangs during install -- &a.scottl;, &a.jeff; Hangs have been reported during the 5.0 snapshot install when installing to mly supported RAID arrays. The same array appears to work fine with 4.7-RELEASE. The card in question is a Mylex AcceleRAID 170, 32 Mb RAM, Firmware revision 7.01-00. This is believed to be the same problem as described in PR 45383.
TrustedBSD MAC Framework userland applications Mostly merged, some minor updates pending &a.rwatson; Most MAC support for user applications and daemons has been merged. Minor updates to support correct labeling in inetd, optional labeling of ttys, and a few other bits and pieces are pending.
Port of LOMAC to TrustedBSD MAC Framework Merge in progress &a.green;,&a.rwatson; New mac_lomac implementation almost completely merged: currently missing is the documentation.
devdPrototype committed. &a.imp;A devd prototype has been committed; will mature as we approach the 5.0 release, including adopting a mature configuration format.
Update Handbook for 5.0 installation In progress &a.trhodes; The handbook largely describes a 4.x installation, and doesn't mention the driver floppy or other recent events. Likewise, it probably hasn't been updated for label editor changes. It also probably talks about the in-kernel ISA configuration editor, which is no longer present. It should talk about ACPI more. Also, the recommended minimum / size should be increased to 128Mb. The handbook should be updated once sysinstall has stabilized, which it should have after RC1.
Further removal of performance degrading debug options In progress re Need to set malloc_junk to 0 in libc. Also need to investigate the use of the -D_LOCK_DEBUG and -D_PTHREADS_INVARIANTS compile flags in libc_r.
Teach sysinstall about rpcbind. -- -- Need to convert references of portmap to rpcbind and add references to rpc.lockd and rpc.statd
Investigate sysinstall problem reports -- -- There are reported problems with sysinstall on i386 not working when the user specifies something other that slice 1 during labeling. More details are forthcoming.
UFS2 bootable on i386Patch committed to HEAD; merge pending testing.&a.imp;, &a.peter;Support for booting UFS2 on non-64-bit platforms required an - increase in the size of the boot blocks due to the size of the - emulated 64-bit arithmetic instructions. In later revisions - of the boot code, we've managed to reduce that size, but as of - RC2, those changes had not been merged to the release branch. - As a result, in RC1 and RC2, UFS2 may only be used for non-root - file systems, not the root file system on 32-bit platforms. On - 64-bit platforms, no such limit exists. We're currently - exploring merging the changes to the boot block to support UFS2 - as the boot file system on all platforms. -

Must Resolve Issues for 5.0-RELEASE

IssueStatusResponsibleDescription
TrustedBSD MAC Framework documentation FreeBSD documentation CVS HEAD: in progress &a.chris;User and developer documentation of the new extensible security framework. Includes guides for live deployment, development of policy modules, and implementation information.

Desired Features for 5.0-RELEASE

IssueStatusResponsibleDescription
User documentation for ACPI. In progress &a.trhodes; The new ACPI code is a very important component of FreeBSD 5.0-RELEASE, but the documentation is insufficient. The manual page does not describe how to perform the basic ACPI tasks. The Usenix paper, man pages, and mailing list archives should be used to write a substantial new section to the FreeBSD Handbook. New users read the Handbook. See PR 43528.
a.out toolchain port -- -- A.out support in the base system was reduced; the removed build chain components should be made into a port.
Rename of local -> unix in rpc socket names Patches committed to HEAD; merge pending testing and review &a.mbr; mbr has requested that we change from using "local" to "unix" for tirpc unix domain socket transport naming, to sync to NetBSD.
NETNCP and nwfs broken -- -- NETNCP and nwfs appear to be as-yet unadapted for KSE, and hence not working. The build needs fixing (probably not all that hard) and then it needs to be tested (harder: few people appear to have Netware servers).
No docs for FAST IPSEC -- -- The new "fast ipsec" implementation has insufficient documentation (man pages, release notes), and this would be highly desirable to have before 5.0-RELEASE.
live file system boot0cfg -- -- boot0cfg -s on a live file system now no longer functions. It would be highly desirable for this, and related functionality, to work.
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.
Documentation rottage: S/Key -- -- We now use OPIE instead of S/Key. The handbook doesn't know that yet.
Documentation wanted: gdbe -- &.chris; 5.0 includes GDBE, supporting encryption of swap and storage. It would be nice if that was documented in the handbook.
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. Direct access to real iir hardware for SCSI developers would be very helpful in resolving this problem.

Areas requiring immediate testing due to 5.0 changes

IssueStatusResponsibleDescription
if_ie, if_le, if_wl network device drivers -- &a.sam; Changes made to the ethernet driver interfaces required sweeping API changes across all ethernet device drivers. While these changes are largely mechanical, they were not tested for three less frequently used interface drivers: if_ie, if_le, and if_wl. They each require testing (and probably fixing) prior to 5.0-RELEASE. If you have access to devices that use these drivers, please perform local testing and provide feedback to Sam Leffler.
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.
inpcb panic involving IP stack sleep -- -- Report of a panic in the TCP syncache thread under load, due to another thread sleeping while holding an inpcb mutex, when not running with WITNESS (WITNESS would have caught the sleep earlier).

Known Issues with 5.0R-RC1 / i386

IssueStatusResponsibleDescription
KDE does not install from ISO In Progress &a.murray; The kde meta-package is not included on the ISO, but many of the component kde packages are, so sysinstall gets confused and installs kde-icontheme instead (more correct behavior would be to fall back to kdebase if kde package is not present).
&footer;