Syscall level |
4217
|
kern
| the kernel's handling of group permissions is suboptimal |
12751
|
kern
| utimes(2) not working when user != fileowner |
12846
|
kern
| brk(2) succeeds on values that should fail. |
14721
|
kern
| It's possible to panic the system by opening a large number of files |
15327
|
kern
| getrusage(2) not reporting all statistics |
20108
|
kern
| KERN_LWP sysctl documentation missing |
25118
|
kern
| semop() fails with EINVAL even when SEMUME is increased |
30115
|
kern
| getrusage returns bogus ru_utime values |
36328
|
kern
| clone(2) with CLONE_FILES can leak POSIX locks |
37427
|
kern
| document _ksem_* syscalls |
38497
|
kern
| Out of memory allocating ksiginfo |
38670
|
kern
| ^Z does not seems to suspend programs that vfork'ed and wait. |
38750
|
kern
| SCHED_FIFO behaviour does not match POSIX |
38965
|
kern
| ktrace locking issue |
38979
|
kern
| Document syscalls of src/sys/kern/sys_sched.c |
39000
|
kern
| sigpending doesn't modify its set argument |
39244
|
kern
| ioctl(2) doesn't describe domains of ioctl commands |
39327
|
kern
| incoherent socket man pages |
39935
|
kern
| ksymsread can leak contents of kernel memory to userspace |
41225
|
kern
| sys_mqueue.c mq->mq_notify_proc can disappear |
42350
|
kern
| frag size in "df -G" output has never been the right value, etc. |
42496
|
kern
| DIOCGPART must die |
42724
|
kern
| select(2) and poll(2) can return non-error status on bad file descriptors |
43025
|
kern
| Non-atomic access to file_t causing corruption |
43587
|
kern
| stupid error message during boot from sysctl if no COMPAT40 in kernel |
44768
|
kern
| sched_rr_get_interval(3) should fail with invalid PID |
45777
|
kern
| two sets of extattr/xattr syscalls |
47439
|
kern
| open() does not check for invalid flags |
48910
|
kern
| msync(2) should return ENOMEM on unmapped pages per POSIX; test criteria in t_msync incorrect |
48962
|
kern
| NetBSD doesn't conform to POSIX with mlock(2) and per-process limits; returns EAGAIN instead of ENOMEM |
44378
|
rump
| uprintf() get lost in rump kernel |
44845
|
rump
| RUMP_MEMLIMIT crashes rump_ffs |
45099
|
rump
| rump rename does not work for syscalls with different args |
Processes/LWPs |
38719
|
kern
| arm, m68k, vax have no cpu_info::ci_curlwp |
38720
|
kern
| curlwp/curcpu should be functions if _MODULE or _LKM |
38751
|
kern
| sched_rr_get_interval() doesn't tell the truth |
39420
|
kern
| stopped processes can hold locks |
39913
|
kern
| exec, fork, exit hooks need locking |
42663
|
kern
| lwp_exit() might race with lwp_free() |
Time |
10102
|
time
| pps api clear/assert inverted |
15688
|
time
| p_stats->p_cru.ru_utime.tv_sec may become negative |
16116
|
time
| clock_getres(CLOCK_REALTIME) is pessimistic |
18170
|
time
| caltontp() function in NTP subsystem ignores "seconds" field of input argument |
24718
|
time
| rtc_offset changes do not fix up process start time |
25637
|
time
| route expiration shouldn't use time(9) |
26354
|
time
| negative running time for process in ps(1) and top(1) output |
32035
|
time
| 3.0 MP machines can't keep time on busy nameservers |
37726
|
time
| 32-bit overflow in adjtime(2) |
38554
|
time
| kernel condvar wait functions should take an absolute timeout |
38789
|
time
| Interval timer should use uptime, not real time. |
40511
|
time
| negative runtime (maybe only in qemu) |
40612
|
time
| timer_settime(2)'s ovalue is not relative time when the timer was created CLOCK_REALTIME |
42189
|
time
| System clock unstable when using PCC as the timecounter when running under SMP |
43997
|
time
| Kernel timer discrepancies |
45090
|
time
| missing callout_ack in sleepq code |
45145
|
time
| SMP cpu time accounting bug |
45539
|
time
| add support for getrusage(2) memory size statistics |
45626
|
time
| System time does not advance correctly when noatime is specified for /var |
45634
|
time
| hardclock_ticks corner cases in vflushnext() et al |
46094
|
time
| cpu0 timer sometimes far off |
47851
|
time
| kernel panic: callout_softclock(v=0x0) -> nfs_timer(arg=0x0) |
47431
|
time
| nanosleep is more like millisleep |
UVM |
6652
|
uvm
| swapping to a sparse file fails unpredictably. |
12371
|
uvm
| swapctl -U (or -d ...) doesn't update process information |
12451
|
uvm
| UVM always overcommits |
25639
|
uvm
| uvm_vslock / uvm_vsunlock problems |
28374
|
uvm
| vmstat(1) reports more pages freed by page daemon than actually are |
33040
|
uvm
| kernel resource shortage + paging activity -> freeze |
33278
|
uvm
| A process repeatetdly sleeps in uvm_fault on "flt_pmfail2"; pagedaemon is repeatedly woken up and does not help it; and nothing else can run. |
34024
|
uvm
| panic: kernel diagnostic assertion "pg == NULL || pg->uobject == uobj" failed |
35741
|
uvm
| PDPOLICY_CLOCKPRO has no tuning parameters analogous to vm.filemax, etc. |
36019
|
uvm
| uvm_readahead.c:ra_startio() assumes (MAXPHYS & (MAXPHYS - 1)) == 0 |
36475
|
uvm
| uvm_fault() in process exit code |
38270
|
uvm
| spurious OOM due to trylock |
38271
|
uvm
| uvm_reclaimable is not mp-safe |
38362
|
uvm
| pmap_growkernel should able to fail |
40726
|
uvm
| uvm panic when xen guest re-mounts block device as rw |
41765
|
uvm
| kernel panic while allocating swap space pages |
45093
|
uvm
| kernel deadlock between TCP and UVM involving callouts |
45413
|
uvm
| unkillable process, uvmexp_anonget waiting for mtsleep |
45504
|
uvm
| A user large file write on fast sata drives can cause a kernel panic |
45598
|
uvm
| expanding uvmexp_sysctl makes top etc fail |
45718
|
uvm
| processes sometimes get stuck and spin in vm_map (kva fragmentation blocking exec args allocation...) |
45746
|
uvm
| UVM_KMF_COLORMATCH/UVM_FLAG_COLORMATCH problems |
46013
|
uvm
| uvm fault in printf_nolog() on MP Xen instance |
46050
|
uvm
| Reader/writer lock error: rw_vector_enter: locking against myself (in uvm during fork) |
46053
|
uvm
| Diagnostic assertion "oldsize != VSIZENOTSET || pgend > oldsize" failed in uvm/uvm_vnode.c |
46136
|
uvm
| processes get stuck in D under high I/O load (likely related to kmem changes) |
46235
|
uvm
| beating on ftruncate kills the system |
46369
|
uvm
| Change type of variable rv. |
47437
|
uvm
| sometimes boot fails with KASSERT(pmap_tlb_pendcount < ncpu) |
DDB |
4761
|
ddb/i386
| setting any watchpoint results in a fatal page fault |
7513
|
ddb/mac68k
| ddb doesn't work on Macintosh LCII |
10016
|
ddb
| ddb can get stuck in infinite page-faults |
11038
|
ddb
| (dM) ddb.onpanic=0 but still drops into ddb |
11295
|
ddb
| change curproc marker in DDB's ps listing |
14618
|
ddb/arm26
| DDB breakpoints on arm26 behave badly |
19774
|
ddb
| ddb can be entered twice and then you can't get out |
20473
|
ddb
| kernel often panics in ddb |
37851
|
ddb
| ddb x/{b,h,l,L} command not working and x/m prints bigendian |
38873
|
ddb
| ddb(4) 'next' command is noisy and slow |
38969
|
ddb/amd64
| ddb backtrace on amd64 doesn't show function arguments |
39388
|
ddb/mac68k
| NetBSD/mac68k DDB uselessness |
40299
|
ddb/alpha
| "machine reboot" doesn't exist on alpha |
40680
|
ddb
| Panic when using whatis <address> from DDB on non-modular port |
40702
|
ddb
| ddb freezes when stepping over lock prefix |
41705
|
ddb
| after continuing from a breakpoint in DDB the system receives a fatal supervisor mode trap |
44242
|
ddb
| Minor but useful improvement to text output from NetBSD debugger |
45647
|
ddb
| kill from ddb broken |
45770
|
ddb
| there should be a ddb command to wake up a lwp |
46240
|
ddb
| options(4) and/or ddb(4) should mention ddb_vgapost |
48256
|
ddb
| ddb manual missing description for DDB_FROMCONSOLE |
48446
|
ddb
| 'boot -d' doesn't enter ddb |
16407
|
kgdb
| remote kgdb `next' command doesn't work in some situation. |
42992
|
kgdb
| KGDB does not work once interrupts are enabled |
Miscellaneous |
2512
|
kern
| fuword() definition problem |
3840
|
kern
| Race condition in the kernel log() and printf() functions |
6434
|
kern
| compress/nocompress minor number bit for tapes |
9449
|
kern
| rasops code is rather endianness-dependant |
16001
|
kern
| Kernel dump files are not compressed on the fly when written to swap |
22616
|
kern
| bootable CD's don't fully identify devices |
31966
|
kern
| problem with scheduler statistics formulation on MP machines |
38241
|
kern
| boot -a ... kernel_lock spinout |
38330
|
kern
| running a -current kernel with lockdebug causes a panic |
38501
|
kern
| NetBSD 4.99.60 (GENERIC) kernel does not handle shutdown -r properly on my hardware |
38564
|
kern
| comcnpollc() is unnecessary |
38565
|
kern
| console outputs CRLF in the "wrong" order |
38583
|
kern
| Kernel objects that are too large to be in wired memory |
38599
|
kern
| panic crashes in early bootstrap |
38608
|
kern
| kernel crash, uvm_fault_internal |
38619
|
kern
| Possible context switch / benchmark improvements |
38779
|
kern
| NetBSD 4 KP : uvm_fault / proc_trampoline |
38801
|
kern
| most drivers assume seralized d_open/d_close |
38888
|
kern
| cnputc() has no locking |
38937
|
kern
| aprint_debug calls make no sense during shutdown |
39008
|
kern
| not all disks support DIOCGSTRATEGY/DIOCSSTRATEGY |
39037
|
kern
| driver(9) does not describe about CFATTACH_DECL_NEW |
39106
|
kern
| patch submissions for http://www.netbsd.org/contrib/projects.html#aprint |
40505
|
kern
| delay(9) is a MD non-interface |
40552
|
kern
| Early kernel panic with POOL_DIAGNOSTIC option |
41243
|
kern
| kernel core dump produces unusable core |
45393
|
kern
| core dumps are unilaterally prevented by unmounted cwd or MNT_NOCOREDUMP even if corename will be valid |
42685
|
kqueue
| waiting a kqueue concurrently may cause kernel panic |
45057
|
kqueue
| kqueue_scan race |
46248
|
kqueue
| kevent won't return when kqueue is being closed |
46523
|
kqueue
| kqueue descriptors can not be passed via SCM_RIGHTS anx. messages |
15837
|
modules
| Kernel should log loading of all loadable modules |
39462
|
modules
| Can't load modules through the bootloader |
41195
|
modules
| NetBSD 5 RC3 can't load kernel modules |
41215
|
modules
| cannot load miniroot as kmod, kernel heap full |
42640
|
modules
| Kernel multiboot implementation lacks support for modules |
43002
|
modules
| required>0 module autounload timestamp race |
45125
|
modules
| Kernel object loader can't detect duplication of global symbol |
45250
|
modules
| Builtin modules cannot be unloaded |
46041
|
modules
| Kernel modules don't work (missing symbol uvmexp_pagesize) |
46734
|
modules
| pool crash on module unload |
46820
|
modules
| Wrong include paths for modules. |
46971
|
modules
| I cannot load uatp module |
48536
|
modules
| Modules need a ref-count mechanism external to the module |
27315
|
pacct
| acct file should be explicitly aligned |
37992
|
PaX
| There's no way to save PaX flags on non-native binaries |
41875
|
PaX
| pax_aslr_elf() use wrong lsb number |
41877
|
PaX
| PaX ASLR reduces available stack size |
46476
|
PaX
| on i386, gdb silently fails to set a breakpoint on PaX-MPROTECTed binary |
32038
|
pipes
| kernel panic in pipe code |
44402
|
pipes
| page fault in supervisor mode in netbsd:filt_piperead |
46275
|
pipes
| Kernel panics on heavy use, kevent / pipes related? |
15174
|
ptrace
| gdb "kill" command does not free sockets |
25529
|
ptrace
| With a MP kernel, ptrace() behaves inconsistently |
44432
|
ptrace
| cannot single-step over a call to sigprocmask() |
40576
|
veriexec
| veriexecgen sets no explicit access mode for some files and the default behaviour of veriexec in face of this cause errors |
41251
|
veriexec
| veriexec locking seems broken |