Commit Graph

76157 Commits

Author SHA1 Message Date
Edward Tomasz Napierala
48cd7df296 Fix array overflow. This routine is only called from procfs,
which is not mounted by default, and I've been unable to trigger
a panic without this fix applied anyway.

Reviewed by:	kib, cperciva
2010-01-24 12:13:38 +00:00
Remko Lodder
b6a18bc939 Add support for the NM10 chipset.
PR:		kern/143118
Submitted by:	Koshin sumitani <ksumitani at mui dot biglobe dot ne dot jp>
Approved by:	imp (mentor, implicit)
MFC after:	3 days
2010-01-24 10:50:20 +00:00
Bjoern A. Zeeb
681ffdf935 Correct a typo.
Submitted by:	kensmith
MFC after:	3 days
2010-01-24 10:22:39 +00:00
Oleksandr Tymoshenko
48fc5de116 - Introduce kernel_kseg0_end variable that marks first address in KSEG0
available for use. All data below this address considered to be used
    by kernel. Along with kernel own data it might be symbol tables
    prepeared by trampoline code, boot loader service data passed for
    further analysis by kernel, etc... By default kernel_kseg0_end points
    to the end of loaded kernel.

- Introduce mips_postboot_fixup function. It checks for symbol information
    copied by ELF trampoline and passes it to KDB
2010-01-24 03:10:48 +00:00
Oleksandr Tymoshenko
2a3e287f9f - Copy symbol-related tables (.symtab and .strtab) to the end of
relocated kernel. We use magic number to signal kernel that
    symbol data is present.
2010-01-24 02:59:22 +00:00
Randall Stewart
50633289e3 Changes the order of the setting the int happened (inside
the lock).
2010-01-24 01:06:02 +00:00
Marcel Moolenaar
5111f97cde Remove cpu_boot() and call efi_reset_system() directly from
cpu_reset().
2010-01-23 23:16:50 +00:00
Marius Strobl
9251d56f5f On LP64 struct ifid is 64-bit aligned while struct fid is 32-bit aligned
so on architectures with strict alignment requirements we can't just simply
cast the latter to the former but need to copy it bytewise instead.

PR:		143010
MFC after:	3 days
2010-01-23 22:38:01 +00:00
Marius Strobl
29b6820f87 Merge r202882 from amd64/i386:
For PT_TO_SCE stop that stops the ptraced process upon syscall entry,
syscall arguments are collected before ptracestop() is called. As a
consequence, debugger cannot modify syscall or its arguments.

In syscall(), reread syscall number and arguments after ptracestop(),
if debugger modified anything in the process environment. Since procfs
stopevent requires number of syscall arguments in p_xstat, this cannot
be solved by moving stop/trace point before argument fetching.

Move the code to read arguments into separate function
fetch_syscall_args() to avoid code duplication. Note that ktrace point
for modified syscall is intentionally recorded twice, once with original
arguments, and second time with the arguments set by debugger.

PT_TO_SCX stop is executed after cpu_syscall_set_retval() already.

Reviewed by:	kib
2010-01-23 22:11:18 +00:00
Joerg Wunsch
ce3853023f Fix breakage introduced to the tnt4882 driver in r202870. This PCI
frontend uses the same uPD7210 backend as the pcii ISA frontend, so
the backend has to cope with both situations.

Also, hide the first printf in pcii_probe (address mismatch) behind
bootverbose as the ISA bus parent tries to probe all configured ISA
devices against each driver, so a the console has been cluttered with
this message for a bunch of unrelated driver probes.

MFC after:	3 days
2010-01-23 21:33:33 +00:00
Alan Cox
040a1eeab2 Simplify the mapping of the system message buffer. Use the direct map just
like ia64 does.
2010-01-23 20:28:37 +00:00
Alan Cox
cf3508519c Handle a race between pmap_kextract() and pmap_promote_pde(). This race is
known to cause a kernel crash in ZFS on i386 when superpage promotion is
enabled.

Tested by:	netchild
MFC after:	1 week
2010-01-23 18:42:28 +00:00
Attilio Rao
b0b9dee5c9 - Fix a race in sched_switch() of sched_4bsd.
In the case of the thread being on a sleepqueue or a turnstile, the
  sched_lock was acquired (without the aid of the td_lock interface) and
  the td_lock was dropped. This was going to break locking rules on other
  threads willing to access to the thread (via the td_lock interface) and
  modify his flags (allowed as long as the container lock was different
  by the one used in sched_switch).
  In order to prevent this situation, while sched_lock is acquired there
  the td_lock gets blocked. [0]
- Merge the ULE's internal function thread_block_switch() into the global
  thread_lock_block() and make the former semantic as the default for
  thread_lock_block(). This means that thread_lock_block() will not
  disable interrupts when called (and consequently thread_unlock_block()
  will not re-enabled them when called). This should be done manually
  when necessary.
  Note, however, that ULE's thread_unblock_switch() is not reaped
  because it does reflect a difference in semantic due in ULE (the
  td_lock may not be necessarilly still blocked_lock when calling this).
  While asymmetric, it does describe a remarkable difference in semantic
  that is good to keep in mind.

[0] Reported by:	Kohji Okuno
			<okuno dot kohji at jp dot panasonic dot com>
Tested by:		Giovanni Trematerra
			<giovanni dot trematerra at gmail dot com>
MFC:			2 weeks
2010-01-23 15:54:21 +00:00
Konstantin Belousov
5b1162b964 For PT_TO_SCE stop that stops the ptraced process upon syscall entry,
syscall arguments are collected before ptracestop() is called. As a
consequence, debugger cannot modify syscall or its arguments.

For i386, amd64 and ia32 on amd64 MD syscall(), reread syscall number
and arguments after ptracestop(), if debugger modified anything in the
process environment. Since procfs stopeven requires number of syscall
arguments in p_xstat, this cannot be solved by moving stop/trace point
before argument fetching.

Move the code to read arguments into separate function
fetch_syscall_args() to avoid code duplication. Note that ktrace point
for modified syscall is intentionally recorded twice, once with original
arguments, and second time with the arguments set by debugger.

PT_TO_SCX stop is executed after cpu_syscall_set_retval() already.

Reported by:	Ali Polatel <alip exherbo org>
Briefly discussed with:	jhb
MFC after:	3 weeks
2010-01-23 11:45:35 +00:00
Konstantin Belousov
a5799a4f27 Staticise sigqueue manipulation functions used only in kern_sig.c.
MFC after:	1 week
2010-01-23 11:43:30 +00:00
Joerg Wunsch
72821110ac Overhaul of the pcii driver:
. Properly allocate all IO space resources.  These cards scatter their
  IO addresses over a range of 0x1600 bytes, and they require an
  additional address for "special interrupt handling".

. Implement the "special interrupt handling" per the GPIB-PCIIA
  Technical Reference Manual; this was apparently not declared for the
  clone card this driver has been originally implemented for, but it
  turned out to be needed for both, an original NI brand PCII/PCIIA
  card as well as the Axiom AX5488 clone.

. Add some diagnostic messages for various resource allocation etc.
  failures during probe.

. Add some comments about the structure of the IO address space that
  is used by these cards.

MFC after:	1 day
2010-01-23 07:54:06 +00:00
Warner Losh
2069a5911e Eliminate octeonregs.h. It was a copy of maltaregs.h with
s/malta/octeon/gi done...
2010-01-23 06:42:47 +00:00
Oleksandr Tymoshenko
6104a13d8b - Move .text and all sections up to .bss inclusive to the beginning of
the script.  Trampoline kernel builder assumes that .text goes right
    after headers and places startup function first in .text segment.
2010-01-23 05:59:10 +00:00
Neel Natu
531c6502cc Remove Sibyte specific code from locore.S that sets the k0seg coherency.
Move it to platform_start() instead.

Approved by: imp (mentor)
2010-01-23 03:19:13 +00:00
Navdeep Parhar
3c0e59de3e Don't forget to release the adapter lock for a no-op. 2010-01-23 01:44:30 +00:00
Warner Losh
8c019fbaa9 Migrate from old "DDB" style debugger to newer KDB style. 2010-01-23 00:24:31 +00:00
Warner Losh
657a57d730 Update from old DDB convetion to initialize debugger to new KDB way.
Always call kdb_init().  If we have KDB enabled, then provide a handy
place to break to the debugger.
2010-01-23 00:18:12 +00:00
Oleksandr Tymoshenko
61b1ecf6d3 - Add driver for PCF2123, SPI real time clock/calendar 2010-01-22 22:14:12 +00:00
Warner Losh
96a33515a3 Declare octeon_get_clock_rate, now exported from octeon_machdep 2010-01-22 20:44:34 +00:00
Warner Losh
659dfb46f1 Remove some irrelevant commented out make options. 2010-01-22 20:42:18 +00:00
Warner Losh
54880c18b4 make note of the nonsensical nature of the values in this hints file. 2010-01-22 20:41:43 +00:00
Warner Losh
f92a1e9b89 o Add support for memory above 256MB on the octeon.
o Force the ebase to be 0x80000000 (the base that we're booted with may
  need to be respected in the future).
o Initialize the clock early so we can initialize the console early
o use panic where we can now use it.
o Tag some code for parsing the boot records as belonging in the cavium sdk.
o remove support for booting on ancient boards...

# we make it further in bootstrapping now: interrupts being enabled in the
# uarts are now taking us out, it seems, for reasons unknown.
2010-01-22 20:40:07 +00:00
Warner Losh
ff9ab61a59 Create a method of last resort for rebooting the mips processor: jump
to the reset vector.  This works for many SoCs where other reset
hardware is either missing or unknown.
2010-01-22 20:32:07 +00:00
Pyun YongHyeon
84e3651eb3 Yukon Ultra2 has 126MHz clock. 2010-01-22 20:15:49 +00:00
Pyun YongHyeon
a91981e4ba s/Mhz/MHz/g
Submitted by:	N.J. Mann <njm <> njm dot me dot uk >
2010-01-22 20:10:12 +00:00
Pyun YongHyeon
d67eba2f3f Use new handshake command for BCM5750 or new controllers. 2010-01-22 18:46:37 +00:00
Pyun YongHyeon
899d684659 Fix a long standing ASF heartbeat sending bug. The initial
implementation of heartbeat interval was 2 but there was typo which
caused the heartbeat is sent approximately every 5 seconds. This
caused unintended controller reset by firmware because firmware
thought OS was crashed.

Submitted by:	Floris Bos < info <> je-eigen-domein dot nl >
Tested by:	Andrzej Tobola < ato <> iem dot pw dot edu dot pl >
2010-01-22 18:35:50 +00:00
Ed Maste
b8c05be0c1 Add H55 ID from Mike Tancsa, with minor rewording from avg@.
PR:		kern/143068
Submitted by:	Mike Tancsa (Sentex)
MFC after:	1 week
2010-01-22 16:05:10 +00:00
Randall Stewart
9b84f07357 This hopefully will fix the network problem I was seeing.
Basically the msg ring interrupt was being re-enabled
inside a spinlock as the thread set it self up for rescheduling.
This won't work since inside the re-enable is another
spin lock.. which means on return from the reenable
the  interrupts have been reenabled. Thus you would
get a clock int and end up panicing holding a spin
lock to long :-o
2010-01-22 14:25:17 +00:00
Rui Paulo
d1f5097233 Remove duplicate bootverbose increment. 2010-01-22 11:52:12 +00:00
Warner Losh
c018866198 Add a suggested improvement. 2010-01-22 09:55:13 +00:00
Alexander Motin
8da30a896a Oops! r202789 broke recording from input mixer. Restore previous "mix"
usage and use "igain" instead for input-to-output monitoring loopback.
2010-01-22 09:54:40 +00:00
Alexander Motin
028b92a574 - Improve tracer, to handle more cases of input-to-output monitoring
loopback.
- Change the meaning of "mix" OSS control. Now it controls loopback level,
according to comments in soundcard.h.
- Allow AD1981HD codecs to use playback mixer. Now driver should be able to
really use it.
- Fix bug in shared muters operation.
2010-01-22 09:31:10 +00:00
Warner Losh
5f04b82bcc Don't clear bss/sbss. The boot loader already does this. In addition,
the Cavium version of the boot loader puts data just after &end, so
our rounding up to the next page in clearing memory overwrote their
data, which meant we'd get a lot of wrong values for parameters to the
system.

While I'm here, remove argc/argv parsing.  Those values aren't passed
in via a0 and a1, so it was a guaranted panic on some boards.
2010-01-22 09:23:34 +00:00
Warner Losh
24e4f4a84d Jump to the mips reset vector. OR in the 0x80000000 for kseg0 and
cast it to a long so it will work on 64-bit targets.

reset now works on octeon, so I don't have to power cycle the board as
often.
2010-01-22 09:19:57 +00:00
Jaakko Heinonen
e48fbf26e8 Truncate read request rather than returning EIO if the request is
larger than MAXPHYS + 1. This fixes a problem with cat(1) when it
uses a large I/O buffer.

Reported by:	Fernando Apesteguía
Suggested by:	jilles
Reviewed by:	des
Approved by:	trasz (mentor)
2010-01-22 08:45:12 +00:00
Michael Tuexen
663fdad84b Use [] instead of [0] for flexible arrays.
Obtained from: Bruce Cran
MFC after: 1 week
2010-01-22 07:53:41 +00:00
Rick Macklem
ff589d0da1 Fix a typo in a comment introduced by r202767.
MFC after:	2 weeks
2010-01-21 21:59:10 +00:00
Jung-uk Kim
d46722a75a Fix a new header inclusion. 2010-01-21 21:31:39 +00:00
Jung-uk Kim
9a179dd8be Merge ACPICA 20100121. 2010-01-21 21:14:28 +00:00
Rick Macklem
f957b30da2 Add a timeout for the negative name cache entries in the NFS client.
This avoids a bogus negative name cache entry from persisting forever
when another client creates an entry with the same name within the
same NFS server time of day clock tick. The mount option negnametimeo
can be used to override the default timeout interval on a
per-mount-point basis. Setting negnametimeo to 0 disables negative
name caching for the mount point.
I also fixed one obvious typo where args.timeo should be
args.maxgrouplist.

Submitted by:	jhb (earlier version)
Reviewed by:	jhb
MFC after:	2 weeks
2010-01-21 20:57:25 +00:00
Alexander Motin
09aa060e71 Print playback channels paths in order of their sequence numbers, not nids. 2010-01-21 09:11:50 +00:00
Neel Natu
b2edbc662e Get rid of redundant setting of interrupt enable bit when restoring the status
register from the PCB.

Remove a couple of misleading comments while I am here. The comments are
misleading because they imply that interrupts will be enabled after the
status register is restored from the PCB. This is not the case because
the processor is at the exception level (SR_EXL is set).

Approved by: imp (mentor)
2010-01-21 02:21:31 +00:00
Oleksandr Tymoshenko
9e3ed0a7fe - Remove unnecessary register writes in activate_device
and deactivate_device
- Save state before attaching driver and restore it when
    detaching
- Clear CLK bit after last bit of byte has been sent over
    the bus providing falling edge for last byte in transfer
- Fix several places where CS0 was always assumed
- Add $FreeBSD$ to ar71xxreg.h
2010-01-21 00:15:59 +00:00
Gabor Kovesdan
b485a59508 - Bump __FreeBSD_version for BSDL bc/dc import to deprecate GNU bc/dc
Approved by:	delphij
2010-01-20 22:45:59 +00:00