John Baldwin [Tue, 8 Apr 2014 21:02:03 +0000 (21:02 +0000)]
Handle single-byte reads from the bvmcons port (0x220) by returning
0xff. Some guests may attempt to read from this port to identify
psuedo-PNP ISA devices. (The ie(4) driver in FreeBSD/i386 is one
example.)
Alexander Motin [Tue, 8 Apr 2014 20:50:48 +0000 (20:50 +0000)]
Add support for SCSI UNMAP commands to CTL.
This patch adds support for three new SCSI commands: UNMAP, WRITE SAME(10)
and WRITE SAME(16). WRITE SAME commands support both normal write mode
and UNMAP flag. To properly report UNMAP capabilities this patch also adds
support for reporting two new VPD pages: Block limits and Logical Block
Provisioning.
UNMAP support can be enabled per-LUN by adding "-o unmap=on" to `ctladm
create` command line or "option unmap on" to lun sections of /etc/ctl.conf.
At this moment UNMAP supported for ramdisks and device-backed block LUNs.
It was tested to work great with ZFS ZVOLs. For file-backed LUNs UNMAP
support is unfortunately missing due to absence of respective VFS KPI.
Sean Bruno [Tue, 8 Apr 2014 20:10:22 +0000 (20:10 +0000)]
Add Stacey Son's binary activation patches that allow remapping of
execution to a emumation program via parsing of ELF header information.
With this kernel module and userland tool, poudriere is able to build
ports packages via the QEMU userland tools (or another emulator program)
in a different architecture chroot, e.g. TARGET=mips TARGET_ARCH=mips
I'm not connecting this to GENERIC for obvious reasons, but this should
allow the kernel module to be built by default and enable the building
of the userland tool (which automatically loads the kernel module).
Marius Strobl [Tue, 8 Apr 2014 07:32:32 +0000 (07:32 +0000)]
Distinguish between the different variants and configurations of Sunix
{MIO,SER}5xxxx chips instead of treating all of them as PUC_PORT_2S.
Among others, this fixes the hang seen when trying to probe the none-
existent second UART on an actually 1-port chip.
Obtained from: NetBSD (BAR layouts)
MFC after: 3 days
Sponsored by: Bally Wulff Games & Entertainment GmbH
Adrian Chadd [Tue, 8 Apr 2014 07:14:14 +0000 (07:14 +0000)]
Add some debugging and forcing of the BAW to match what the current
tracked BAW actually is.
The net80211 code that completes a BAR will set tid->txa_start (the
BAW start) to whatever value was called when sending the BAR.
Now, in case there's bugs in my driver code that cause the BAW
to slip along, we should make sure that the new BAW we start
at is actually what we currently have it at, not what we've sent.
This totally breaks the specification and so this stays a printf().
If it happens then I need to know and fix it.
Whilst here, add some debugging updates:
* add TID logging to places where it's useful;
* use SEQNO().
Adrian Chadd [Tue, 8 Apr 2014 07:08:59 +0000 (07:08 +0000)]
Correct the actual definition of ath_tx_tid_filt_comp_single() to
match how it's used.
This is another bug that led to aggregate traffic hanging because
the BAW tracking stopped being accurate. In this instance, a filtered
frame that exceeded retries would return a non-error, which would
mean the caller would never remove it from the BAW. But it wouldn't
be added to the filtered list, so it would be lost forever. There'd
thus be a hole in the BAW that would never get transmitted and
this leads to a traffic hang.
Adrian Chadd [Tue, 8 Apr 2014 07:00:43 +0000 (07:00 +0000)]
Don't resume a TID on each filtered frame completion - only do it if
we did suspend it.
The whole suspend/resume TID queue thing is supposed to be a matched
reference count - a subsystem (eg addba negotiation, BAR transmission,
filtered frames, etc) is supposed to call pause() once and then resume()
once.
ath_tx_tid_filt_comp_complete() is called upon the completion of any
filtered frame, regardless of whether the driver had aleady seen
a filtered frame and called pause().
So only call resume() if tid->isfiltered = 1, which indicates that
we had called pause() once.
This fixes a seemingly whacked and different problem - traffic hangs.
What was actually going on:
* There'd be some marginal link with crappy behaviour, causing filtered
frames and BAR TXing to occur;
* A BAR TX would occur, setting the new BAW (block-ack window) to seqno n;
* .. and pause() would be called, blocking further transmission;
* A filtered frame completion would occur from the hardware, but with
tid->isfiltered = 0 which indiciates we haven't actually marked
the queue yet as filtered;
* ath_tx_tid_filt_comp_complete() would call resume(), continuing
transmission;
* Some frames would be queued to the hardware, since the TID is now no
longer paused;
* .. and if some make it out and ACked successfully, the new BAW
may be seqno n+1 or more;
* .. then the BAR TX completes and sets the new seqno back to n.
At this point the BAW tracking would be loopy because the BAW
start was modified but the BAW ring buffer wasn't updated in lock
step.
Adrian Chadd [Tue, 8 Apr 2014 02:36:27 +0000 (02:36 +0000)]
Add a basic set of data points which count the number of sleep entries
that are being done by the OS.
For now this'll match up with the "wakeups"; although I'll dig deeper into
this to see if we can determine which sleep state the CPU managed to get
into. Most things I've seen these days only expose up to C2 or C3 via
ACPI even though the CPU goes all the way down to C6 or C7.
Fix panic on load new driver while vt(4) is in VGA textmode.
o Mute terminal while vt(4) driver change in progress.
o Reset VDF_TEXTMODE before init new driver.
o Assign default font, if new driver is not in TEXTMODE.
o Do not update screen while driver changing.
Resolved by: adrian
Reported by: tyler
MFC after: 7 days
Sponsored by: The FreeBSD Foundation
Loosen the processing of *_IF_aliasN vars to be less strict. Previously,
the first alias had to be _alias0 and processing stopped at the first non-
defined variable (preventing gaps). Allowing gaps gives the administrator
the ability to group aliases in an adhoc manner and also lifts the
requirement to renumber aliases simply to comment-out an existing one.
Aliases are processed in numerical ascending order.
For GNU as, add two missing modes for each of the fcomip and fucomip
instructions. Partially obtained from OpenBSD by Pedro Giffuni, while I
added the fcomip variants.
Apparently this should help with compiling certain variants of WebKit.
Ed Schouten [Mon, 7 Apr 2014 21:11:29 +0000 (21:11 +0000)]
Clean up shutdown_nice(). Just send the right signal to init(8).
Right now, init(8) cannot distinguish between an ACPI power button press
or a Ctrl+Alt+Del sequence on the keyboard. This is because
shutdown_nice() sends SIGINT to init(8) unconditionally, but later
modifies the arguments to reboot(2) to force a certain behaviour.
Instead of doing this, patch up the code to just forward the appropriate
signal to userspace. SIGUSR1 and SIGUSR2 can already be used to halt the
system.
While there, move waittime to the function where it's used; kern_reboot().
- Don't include sys/caprights.h, leverage the fact that cap_rights_t
is also defined in sys/types.h.
- Include sys/types.h directly.
- For systems that do not have cap_rights_t, define it, so we can use
it in au_to_rights() prototype.
Implement the final missing sysctls by moving ipf_auth_softc_t from
ip_auth.c to ip_auth.h. ip_frag_soft_t moves from ip_frag.c to
ip_frag.h. mlfk_ipl.c creates sysctl MIBs that reference control blocks
that are dynamically created when IP Filter is loaded. This necessitated
creating them on-the-fly rather than statically at compile time.
Ed Schouten [Mon, 7 Apr 2014 18:10:49 +0000 (18:10 +0000)]
Implement kqueue(2) for procdesc(4).
kqueue(2) already supports EVFILT_PROC. Add an EVFILT_PROCDESC that
behaves the same, but operates on a procdesc(4) instead. Only implement
NOTE_EXIT for now. The nice thing about NOTE_EXIT is that it also
returns the exit status of the process, meaning that we can now obtain
this value, even if pdwait4(2) is still unimplemented.
Notes:
- Simply reuse EVFILT_NETDEV for EVFILT_PROCDESC. As both of these will
be used on totally different descriptor types, this should not clash.
- Let procdesc_kqops_event() reuse the same structure as filt_proc().
The only difference is that procdesc_kqops_event() should also be able
to deal with the case where the process was already terminated after
registration. Simply test this when hint == 0.
- Fix some style(9) issues in filt_proc() to keep it consistent with the
newly added procdesc_kqops_event().
- Save the exit status of the process in pd->pd_xstat, as we cannot pick
up the proctree_lock from within procdesc_kqops_event().
The units program is likely little used. It is even less likely that a script
will want the units program to print out its version number by passing -v.
GNU units uses -V for version and -v for verbosity.
Increase compatibility between these two versions (written by the same author)
by switching our flag as well.
Take this opportunity to remove bogus information about the version number and
just call it 'FreeBSD units'.
Ed Maste [Mon, 7 Apr 2014 00:49:15 +0000 (00:49 +0000)]
Do not build the amd64 UEFI loader with GCC
The UEFI loader causes buildworld to fail when building with (in-tree)
GCC, due to a typedef redefinition. As it happens the in-tree GCC
cannot successfully build the UEFI loader anyhow, as it does not support
__attribute__((ms_abi)). Thus, just avoid trying to build it with GCC, rather than disconnecting it from the build until the underlying issue
is fixed.
fts_read() leaves errno unchanged on EOF and sets it on error, so set errno
to 0 before calling it. Also, don't trust finish_execplus() to leave errno
unchanged.
Ed Schouten [Sun, 6 Apr 2014 20:00:42 +0000 (20:00 +0000)]
Nit: fix locking of p->p_state in procdesc_close().
According to <sys/proc.h>, this field needs to be locked with either the
p_mtx or the p_slock. In this case the damage was quite small. Instead
of being reaped, the process would just be reparented to init, so it
could be reaped from there.
An all-or-nothing approach to labels isn't flexible enough. Embedded
systems need fine-grained control over what's in and what's out.
That's ideal. For now, separate GPT labels from the rest and allow
g_label to be built with just GPT labels.
Make sure we don't free memory that's already been freed by setting
the geom->softc pounter to NULL before freeing the g_slicer softc.
In g_slicer_free() the pointer is checked first.
Ian Lepore [Sun, 6 Apr 2014 00:17:41 +0000 (00:17 +0000)]
Add a couple more required TLB flushes.
These should have been part of r264129, they are part of the overall set
of changes that got several weeks of testing. I must have fumbled them
while merging various patchsets.
Warner Losh [Sat, 5 Apr 2014 22:43:23 +0000 (22:43 +0000)]
Make the vmm code compile with gcc too. Not entirely sure things are
correct for the pirbase test (since I'd have thought we'd need to do
something even when the offset is 0 and that test looks like a
misguided attempt to not use an uninitialized variable), but it is at
least the same as today.
Do not prevent processes from making changes to the baudrate or the
CLOCAL and HUPCL control flags. There are legit reasons for allowing
those to be changed. When /etc/ttys has the "3wire" type (without a
baudrate) for the serial port that is the low-level console, then
this change has no effect.
In freebsd32_sendmsg(), replace the call to sockargs() followed by a
call to freebsd32_convert_msg_in() with freebsd32_copyin_control() to
readin and convert in a single step. This makes it simpler to put all
the control messages in a single mbuf or mbuf cluster as per the
limitations imposed upon us by ip6_setpktopts().
The logic is as follows:
1. Go over the array of control messages to determine overall size
and include extra padding for proper alignment as we go.
2. Get a mbuf or mbuf cluster as needed or fail if the overall
(adjusted) size is larger than a cluster.
3. Go over the array of control messages again, but now copy them
into kernel space and into aligned offsets.
4. Update the length of the control message to take padding between
the header and the data into account (but not for padding added
between one control message and the next).
Obtained from: Juniper Networks, Inc.
MFC after: 1 week
Accept RFC 2292 option values so that RFC 2292 compliant programs that
are unaware of RFC 3542 can construct control messages.
The kernel disallows mixing RFC 2292 behaviour with RFC 3542 behaviour.
Only sockets that have specifically been marked as using the RFC 2292
API can use RFC 2292 specific options. This is all good and well, but
libc itself seems inconsistent with this.
The root cause of this inconsistency seems to relate to the definitions
of IPV6_HOPOPTS and IPV6_DSTOPTS. They are defined in RFC 2292 and re-used
in RFC 3542, yet have distinct values in the kernel. It's for this reason
that the kernel also has definitions for IPV6_2292HOPOPTS and
IPV6_2292DSTOPTS. Not so in libc.
For example: some program calls inet6_option_init() (defined by RFC 2292)
with the RFC 2292 defined IPV6_HOPOPTS and IPV6_DSTOPTS. Before RFC 3542,
this was translated to values of 22 and 23 (resp.) The libc implementation
correctly checks that only options IPV6_HOPOPTS and IPV6_DSTOPTS are given
(as per RFC 2292) but since these defines have taken on the values defined
by RFC 3542 (values 49 and 50 resp,) rejects the correct option values
(22 and 23) passed said program and returns -1.
The precisie fix is to have inet6_option_init() and friends only accept the
RFC 2292 defined IPV6_HOPOPTS & IPV6_DSTOPTS, but that breaks other code
(like mld6query(8)), which seem to not be aware of RFC 3542 and how it
hi-jacked the option names. So the best fix is to accept the options from
both.
Obtained from: Juniper Networks, Inc.
MFC after: 1 week
The getlogin_basic() function can return a 0 status with a NULL
pointer for the login name (result). Make sure to handle that
case properly. Improve robustness by checking namelen and then
nul-terminating the provided buffer to simplify subsequent logic.
Obtained from: Juniper Networks, Inc.
MFC after: 1 week
Warner Losh [Sat, 5 Apr 2014 17:54:36 +0000 (17:54 +0000)]
The proper way to request no man pages currently is NO_MAN=xxx. Use it
in preference to the user WITHOUT_MAN knob, which should never be set
in normal src Makefiles.
- Fix the setup of interrupts for banks 2 and 3 on AM335x.
On AM335x each one of the four GPIO banks has two physical interrupt
lines, so we now allocate resources and setup our interrupt handler for
all the (8) available interrupts.
On OMAP3 and OMAP4 there is only one interrupt for each GPIO bank (6
banks, 6 interrupts), but there are two set of registers where the
first one is used to setup the delivery of interrupts to the MPU and
the second set, setup the delivery of interrupts to the DSP.
On AM335x, each set of registers controls each one of the interrupt
lines.
- Remove nonexistent registers for OMAP4 and AM335x, replace their use with
the correct ones for these SoCs.
Ian Lepore [Sat, 5 Apr 2014 16:08:13 +0000 (16:08 +0000)]
Add ioctl(2) calls to uftdi(4) to access bitbang, MPSSE, CPU_FIFO, and
other modes supported by the FTDI serial adapter chips.
In addition to adding the new ioctls, this change removes all the code
that reset the chip at attach and open/close time, and also the code
that turned on RTS/CTS flow control on open without any permission to do
so (that was just always a bug in the driver).
When FTDI chips are configured as GPIO or MPSSE or other special-purpose
uses by an attached serial eeprom, the chip will power on with certain
pins driven or floating, and it's important that the driver not do
anything to the chip to perturb that unless it receives a specific
command to do so. When used for "plain old serial comms" the chip
powers on into the right mode and never needs to be reset while it's
running to operate properly, so this change is transparent to most users.