Provide a sysctl switch that allows to install ARP entries
with multicast bit set. FreeBSD refuses to install such
entries since 9.0, and this broke installations running
Microsoft NLB, which are violating standards.
ray [Mon, 3 Sep 2012 08:52:05 +0000 (08:52 +0000)]
Add kern.hintmode sysctl variable to show current state of hints:
0 - loader hints in environment only;
1 - static hints only
2 - fallback mode (Dynamic KENV with fallback to kernel environment)
Add kern.hintmode write handler, accept only value 2. That will switch
static KENV to dynamic. So it will be possible to change device hints.
dim [Sun, 2 Sep 2012 14:46:18 +0000 (14:46 +0000)]
Partially revert r239959, after actually fixing most of the clang
warnings in sys/gnu/fs/xfs. The only warnings that still need to be
suppressed are those about array bound overruns of flexible array
members in xfs_dir2_{block,sf}.c, which are too expensive (in terms of
cascading code changes) to fix.
In soreceive_generic() remove the optimization for the case when
MSG_WAITALL is set, and it is possible to do the entire receive
operation at once if we block (resid <= hiwat). Actually it might make
the recv(2) with MSG_WAITALL flag get stuck when there is enough space
in the receiver buffer to satisfy the request but not enough to open
the window closed previously due to the buffer being full.
The issue can be reproduced using the following scenario:
On the sender side do 2 send(2) requests:
1) data of size much smaller than SOBUF_SIZE (e.g. SOBUF_SIZE / 10);
2) data of size equal to SOBUF_SIZE.
On the receiver side do 2 recv(2) requests with MSG_WAITALL flag set:
1) recv() data of SOBUF_SIZE / 10 size;
2) recv() data of SOBUF_SIZE size;
We totally fill the receiver buffer with one SOBUF_SIZE/10 size request
and partial SOBUF_SIZE request. When the first request is processed we
get SOBUF_SIZE/10 free space. It is just enough to receive the rest of
bytes for the second request, and soreceive_generic() blocks in the
part that is a subject of this change waiting for the rest. But the
window was closed when the buffer was filled and to avoid silly window
syndrome it opens only when available space is larger than sb_hiwat/4
or maxseg. So it is stuck and pending data is only sent via TCP window
probes.
In soreceive_generic() when checking if the type of mbuf has changed
check it for MT_CONTROL type too, otherwise the assertion
"m->m_type == MT_DATA" below may be triggered by the following scenario:
- the sender sends some data (MT_DATA) and then a file descriptor
(MT_CONTROL);
- the receiver calls recv(2) with a MSG_WAITALL asking for data larger
than the receive buffer (uio_resid > hiwat).
adrian [Sun, 2 Sep 2012 04:56:29 +0000 (04:56 +0000)]
AR5212 radar pulse fixes.
Fix the strong signal diversity capability setting - I had totally
messed up the indentation.
Set the default values to match what's in the .ini for now, rather than
what values I had previously gleaned from places. This seems to work
quite well for the early AR5212 NICs I have. Of course, later NICs
have different PHYs and the radar configuration is very card/board
dependent..
Rework all non-contributed files that use `struct timezone'.
This structure is not part of POSIX. According to POSIX, gettimeofday()
has the following prototype:
int gettimeofday(struct timeval *restrict tp, void *restrict tzp);
Also, POSIX states that gettimeofday() shall return 0 (as long as tzp is
not used). Remove dead error handling code. Also use NULL for a
nul-pointer instead of integer 0.
While there, change all pieces of code that only use tv_sec to use
time(3), as this provides less overhead.
Fix panic in procdesc that can be triggered in the following scenario:
1. Process A pdfork(2)s process B.
2. Process A passes process descriptor of B to unrelated process C.
3. Hit CTRL+C to terminate process A. Process B is also terminated
with SIGINT.
4. init(8) collects status of process B.
5. Process C closes process descriptor associated with process B.
When we have such order of events, init(8), by collecting status of
process B, will call procdesc_reap(). This function sets pd_proc to NULL.
Now when process C calls close on this process descriptor,
procdesc_close() is called. Unfortunately procdesc_close() assumes that
pd_proc points at a valid proc structure, but it was set to NULL earlier,
so the kernel panics.
The patch also adds setting 'p->p_procdesc' to NULL in procdesc_reap(),
which I think should be done.
After I made the previous commit, I noticed the code does some things it
shouldn't. It casts a struct timeval to a time_t, assuming tv_sec is the
first member. Also, we are not interested in microseconds, so it is
better to just call time(NULL).
adrian [Sat, 1 Sep 2012 05:43:30 +0000 (05:43 +0000)]
Fix the PHY / CRC error bug in the AR5212 HAL, which apparently also pops
up on (at least) the AR5413.
The 30 second summary - if a CRC error frame comes in during PHY error
processing, that CRC bit will be set for all subsequent frames until
a non-CRC error frame is processed.
So to allow for accurate PHY error processing (Radar, and ANI on the AR5212
HAL chips) just tag the frame as being both CRC and PHY - let the driver
decide what to do with it.
Bring some changes from Bull's NFSv4 libtirpc implementation.
We especifically ignored the glibc compatibility changes
but this should help interaction with Solaris and Linux.
____
Fixed infinite loop in svc_run()
author Steve Dickson
Tue, 10 Jun 2008 12:35:52 -0500 (13:35 -0400)
Fixed infinite loop in svc_run()
____
__rpc_taddr2uaddr_af() assumes the netbuf to always have a
non-zero data. This is a bad assumption and can lead to a
seg-fault. This patch adds a check for zero length and returns
NULL when found.
author Steve Dickson
Mon, 27 Oct 2008 11:46:54 -0500 (12:46 -0400)
____
Changed clnt_spcreateerror() to return clearer
and more concise error messages.
author Steve Dickson
Thu, 20 Nov 2008 08:55:31 -0500 (08:55 -0500)
____
Converted all uid and gid variables of the type uid_t and gid_t.
author Steve Dickson
Wed, 28 Jan 2009 12:44:46 -0500 (12:44 -0500)
____
libtirpc: set r_netid and r_owner in __rpcb_findaddr_timed
These fields in the rpcbind GETADDR call are being passed uninitialized
to CLNT_CALL. In the case of x86_64 at least, this usually leads to a
segfault. On x86, it sometimes causes segfaults and other times causes
garbage to be sent on the wire.
rpcbind generally ignores the r_owner field for calls that come in over
the wire, so it really doesn't matter what we send in that slot. We just
need to send something. The reference implementation from Sun seems to
send a blank string. Have ours follow suit.
author Jeff Layton
Fri, 13 Mar 2009 11:44:16 -0500 (12:44 -0400)
____
libtirpc: be sure to free cl_netid and cl_tp
When creating a client with clnt_tli_create, it uses strdup to copy
strings for these fields if nconf is passed in. clnt_dg_destroy frees
these strings already. Make sure clnt_vc_destroy frees them in the same
way.
author Jeff Layton
Fri, 13 Mar 2009 11:47:36 -0500 (12:47 -0400)
dim [Fri, 31 Aug 2012 23:28:41 +0000 (23:28 +0000)]
Fix a twelve year old bug in readelf: when process_dynamic_segment()
encounters a DT_RUNPATH entry, the global dynamic_info[] array is
overrun, causing some other global variable to be overwritten.
In my testcase, this was the section_headers variable, leading to
segfaults or jemalloc assertions when it was freed later on.
Thanks to Koop Mast for providing samples of a few "bad" .so files.
ed [Fri, 31 Aug 2012 22:22:14 +0000 (22:22 +0000)]
Properly enable Clang-style atomics when available.
In addition to testing against cxx_atomic, we must check c_atomic. The
former is only set when building C++ code. Also use __has_extension
instead of __has_feature. This allows us to use the atomics outside of
C11.
Reported by: Ariane van der Steldt <ariane stack nl>
PR: threads/170073
dim [Fri, 31 Aug 2012 21:45:49 +0000 (21:45 +0000)]
Work around several warnings from clang in the xfs filesystem, when
linking it statically into the kernel. With our gcc in base there are
no warnings, so also remove the WERROR= from the module makefile.
jhb [Fri, 31 Aug 2012 21:27:23 +0000 (21:27 +0000)]
Similar to how r171350 fixed linking of kernel modules containing
firmware objects by adding --no-warn-mismatch to the linker flags,
add --no-warn-mismatch when linking firmware objects (*.fwo) as
well as to the link of the main kernel file. This permits firmware
modules to be statically linked into an ia64 kernel.
jhb [Fri, 31 Aug 2012 21:10:38 +0000 (21:10 +0000)]
The implied source variable (.IMPSRC) didn't actually work in my previous
commit. Change this to use .ALLSRC instead, but be careful to only use
the .fw file for NORMAL_FWO to ignore opt_global.h.
jhb [Fri, 31 Aug 2012 20:54:30 +0000 (20:54 +0000)]
Add common rules for building firmware object files (NORMAL_FW to run
uudecode, and NORMAL_FWO to use ld to build the .fwo file) and use those
instead of explicit ld/uudecode invocations in sys/conf/files. Apart from
increasing readability, this makes it possible to adjust the flags used for
firmware objects in one place.
ed [Fri, 31 Aug 2012 18:26:39 +0000 (18:26 +0000)]
Relicense ac(8).
Though the license of the original ac(8) source code provides many
liberties, we are already somewhat violating it. The license requires us
to clearly comment any modifications to the source code, as the original
authors of course do not want to get bug reports for modified versions
of ac(8). This is something we have not done consistently.
The need for such a license clause has become less over time. It is very
unlikely that end users will contact the original authors, as the
copyright is from 1994. I contacted both the copyright holders. They
responded in a timely fashion and were more than willing to relicense it
to a 2-clause BSD license. To address any concerns about bug reports
going to the original authors instead of me, add my own name and email
address to the copyright statements as well.
MFC after: 1 month
Special thanks to: Christopher Demetriou and Simon Gerraty
jhb [Fri, 31 Aug 2012 14:35:01 +0000 (14:35 +0000)]
Add a -h flag similar to the -h flag for ln to force mv(1) to treat a
symbolic link to a directory for the target as a symbolic link instead of
a directory. This makes it possible to atomically update a symbolic
link using rename().
scottl [Fri, 31 Aug 2012 10:07:38 +0000 (10:07 +0000)]
Heavily optimize the case of small RX packets of 160 bytes or less. For
this case, allocate a plain mbuf and copy the frame into it, then send the
copy up the stack, leaving the original mbuf+cluster in place in the
receive ring for immediate re-use. This saves a trip through 2 of the
3 zones of the compound mbuf allocator, a trip through busdma, and a trip
through the 1 of the 3 mbuf destructors. For our load at Netflix, this can
lower CPU consumption by as much as 20%. The copy algorithm is based on
investigative work from Luigi Rizzo earlier in the year.
scottl [Fri, 31 Aug 2012 09:42:46 +0000 (09:42 +0000)]
Essentially revert r239912. The amr_periodic function hadn't been armed in
over 10 years and was dead code; the previous revision exposed it as such to
CLANG. The solution is to cull the whole thing.
attilio [Thu, 30 Aug 2012 21:22:47 +0000 (21:22 +0000)]
Post r222812 KTR_CPUMASK started being initialized only as a tunable
handler and not more statically.
Unfortunately, it seems that this is not ideal for new platform bringup
and boot low level development (which needs ktr_cpumask to be effective
before tunables can be setup).
Because of this, add a way to statically initialize cpusets, by passing
an list of initializers, divided by commas. Also, provide a way to enforce
an all-set mask, for above mentioned initializers.
This imposes some differences on how KTR_CPUMASK is setup now as a
kernel option, and in particular this makes the words specifications
backward wrt. what is currently in -CURRENT. In order to avoid mismatches
between KTR_CPUMASK definition and other way to setup the mask
(tunable, sysctl) and to print it, change the ordering how
cpusetobj_print() and cpusetobj_scan() acquire the words belonging
to the set.
Please give a look to sys/conf/NOTES in order to understand how the
new format is supposed to work.
Also, ktr manpages will be updated shortly by gjb which volountereed
for this.
This patch won't be merged because it changes a POLA (at least
from the theoretical standpoint) and this is however a patch that
proves to be effective only in development environments.
jhb [Thu, 30 Aug 2012 17:47:39 +0000 (17:47 +0000)]
Attach interrupt handlers during attach instead of during the first time
the interface is brought up. Without this, the boot time interrupt
round-robin assignment does not think the allocated interrupt resources
are active and leaves them assigned to CPU 0.
While here, add descriptive tags to each interrupt handler when MSI-X
is used.
ed [Thu, 30 Aug 2012 16:45:27 +0000 (16:45 +0000)]
Massively refactor ac(8).
- Use queue(3) -- not some homegrown implementation of linked lists.
- Rename structures to _entry, as they are entries in the linked list --
not the lists themselves.
- Don't store entire copies of struct utmpx in utmpx_entry, but only the
members we're interested in. Large fields such as hostnames are not
needed during the execution of the program.
- Give structure members useful names, instead of `name'.
- While there, use struct timevals instead of time_t's internally. This
is not strictly useful, but while we're at it...
- Mark stuff static.
- Add missing const keywords.
- Remove unneeded prototypes.
- Remove workaround for sparc64-specific utmp problems. These don't
apply to utmpx.
- Don't discard entries when timestamps are not monotone. This shouldn't
ever happen with utmpx, but discarding them is a bit too harsh.
- Remove debug code. We nowadays have `getent utmpx', which can be used
to analyze logfiles in depth.
- Use proper uppercasing/periods in comments.
- Print output of `ac -p' sorted alphabetically, instead of first
occurrence.
- Properly check against pts/* instead of tty[PQRSpqrs]* to determine
whether a TTY is a pseudo-terminal.
hselasky [Thu, 30 Aug 2012 16:19:05 +0000 (16:19 +0000)]
Preparations for adding USB HOST mode to the DWC OTG driver.
Merge register file with external one and put all register
definitions in a separate file.
glebius [Thu, 30 Aug 2012 12:18:45 +0000 (12:18 +0000)]
In ifc_alloc_unit():
- In the !wildcard case, return ENOSPC instead of confusing EEXIST
in case if ifc->ifc_maxunit reached.
- Fix unit leak, that I've introduced in previous revision.
adrian [Wed, 29 Aug 2012 22:58:52 +0000 (22:58 +0000)]
More attempts at space saving.
* add cam as a module to build - but build in scbus/da for now, as
"cam" as a module includes all cam devices. Hardly space saving.
* Don't build FFS snapshot support.
dim [Wed, 29 Aug 2012 18:49:41 +0000 (18:49 +0000)]
Teach libdwarf about the DW_FORM_flag_present dwarf attribute, so
programs using libdwarf (such as ctfconvert) don't error out on files
containing the attribute.
dim [Wed, 29 Aug 2012 18:37:10 +0000 (18:37 +0000)]
Teach gdb about the DW_FORM_flag_present dwarf attribute, so it doesn't
error out on files that contain it. (This attribute can be emitted by
newer versions of clang.)
dim [Wed, 29 Aug 2012 18:22:52 +0000 (18:22 +0000)]
Remove the argument-less .align directive in sys/i386/bios/smapi_bios.S.
Specifying no argument is undocumented in the gas manual, and clang's
integrated assembler refuses to parse it. Also, removing it causes no
change at all in the resulting object file.
adrian [Wed, 29 Aug 2012 16:58:51 +0000 (16:58 +0000)]
There's no nede to allocate a DMA map just before calling bus_dmamem_alloc().
In fact, bus_dmamem_alloc() happily NULLs the dmat pointer passed in,
before replacing it with its own.
This fixes a MIPS crash when kldload'ing if_ath/if_ath_pci -
bus_dmamap_destroy() was passed in a NULL dmat pointer and was doing
all kinds of very bad things.
marius [Wed, 29 Aug 2012 16:56:50 +0000 (16:56 +0000)]
- Unlike cache invalidation and TLB demapping IPIs, reading registers from
other CPUs doesn't require locking so get rid of it. As the latter is used
for the timecounter on certain machine models, using a spin lock in this
case can lead to a deadlock with the upcoming callout(9) rework.
- Merge r134227/r167250 from x86:
Avoid cross-IPI SMP deadlock by using the smp_ipi_mtx spin lock not only
for smp_rendezvous_cpus() but also for the MD cache invalidation and TLB
demapping IPIs.
- Mark some unused function arguments as such.