| Age | Commit message (Collapse) | Author |
|---|
|
From Alexander Yurchenko <grange@rt.mipt.ru>
|
|
From Arla 0.35.5 through David Krause <openbsd@davidkrause.com>
|
|
first char is a space. Patch from sacrificial-spam-address@horizon.com
|
|
in the input buffer. This prevents a rare buffer overflow on very long
header lines where one or more entries has a comment in it but the
entries have no space after the comma *and* the amount of extra space
needed to add a space after each comma is greater than the length of
the comments that will be removed. This is debian bug #108677
2) In skin(), use a temporary variable in the realloc() and don't
die if realloc() fails since its only purpose is to shrink the
buffer, not expand it (and thus is not fatal).
|
|
|
|
|
|
|
|
|
|
|
|
and simpler.
|
|
Allow multiple occurences of "struct foo".
Add syntaxic sugar to make use of "struct" and "member" keywords more
useful in genassym.cf files.
This will be badly documented in a manpage, to come in a few minutes.
|
|
|
|
|
|
|
|
Now uses mii/mii_bitbang.
Driver still has performance issues, It does not transfer
as fast data as it should. No worse than before.
Need to investigate a dhclient reported error:
'ip length 331 disagrees with bytes recieved 336'. only on if_bm.
|
|
Fix from janjaap@stack.nl. PR 1775
|
|
|
|
|
|
kernel. From based on NetBSD patches.
|
|
o handles decode of all 64 32 bit registers (based on type)
o easier to read (at least imo).
o needs some optimization (clarity was the first goal here)
|
|
|
|
|
|
Differentiate between return-icmp and return-icmp6,
icmp-type and ipv6-icmp-type. ok dhartmei@
|
|
|
|
'ICMP error message for bad proto' messages and breaking traceroute etc.
Please increase debugging level (pfctl -x m) while testing.
|
|
as some output format (netmasks, inet/inet6, return-icmp).
|
|
|
|
Verify that regions which were allocated by stealing are not returned
to phys_map, perhaps before phys_map is initialized. Not seen
to be a problem, but is potentially.
Also, the system can deal with not having the devices mapped with BATs.
In fact after some testing the systems seem more stable when BATs are
not used for devices. Stop using them, at least for now.
|
|
?+(64k), instead print 0+0x<address>, at least that is useful.
|
|
interrupt after one second. Originally the interrupt was disabled
permanently after it fired once. This causes futher hot plug/removals
to stop working. However the interrupt will fire several times in succession
as a device is inserted/removed. This effectively debounces the interrupt.
Apparently only some systems actually plug directly into the root hub:
Most PCI cards do not, however newer apple motherboards all do.
This was sent around for review some time back and again recently, this
time with no feedback.
|
|
add a missing initialization of the sym pointer.
Add some functionality which allows a program to open itself dlopen(NULL),
so that it can then look up symbols in the executable itself.
Note that the program can only access exported variables, either by
exporting all variables with the ld option -E or externally referrenced.
Fix bug in dlsym() where it would return failure when looking up symbols.
It was testing the offset of the found symbol, not if the symbol was found.
|
|
(the previous priority didn't help performance in tests on a hacked
up BPF and it weighed down the load average)
|
|
|
|
|
|
|
|
packets and is obviously invalid (and not NULL) for IPv6 packets (hence
crashed). Pass ifp down instead.
sizeof(ih) instead of sizeof(&ih) for pf_pull_hdr() from pf_test6().
|
|
|
|
|
|
|
|
|
|
possible bugs in it which could cause the code to spin indefinately
attempting to remove all mappings for a page.
This is now able to survive a paging death program and additional other
testing.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
w/ bus_dma(9) and rework drivers accordingly.
make drivers use bus_dma as well, except for if_bm (later ;) .
additionally, sync wdc_obio w/ netbsd.
drahn@ ok, tested by miod@, pval@, brad@, mickey@
|
|
|
|
|
