Do not remove logging sockets. This fixes an issue where logging
sockets placed into prisons from the host environment get clobbered
by the prison's instance of cleanvar. (assuming /etc/rc is run in
the prison).
The divide by zero panic must have been due to a bogus
period value. I suppose the BT adapter driver should be
fixed, but more importantly we should protect against
dividing by zero.
peter [Thu, 14 Apr 2005 03:16:58 +0000 (03:16 +0000)]
Allow user processes to completely empty out their LDT, now that user
processes run from segment selectors that live in the GDT. Doing this
used to be equivalent to committing suicide, but now this is a NOP.
peter [Wed, 13 Apr 2005 22:57:17 +0000 (22:57 +0000)]
Change the segment limits to 4GB, we set the user accessible bit on all
of the kernel address space already. Intel recommend this anyway, because
using a non-4GB limit adds an additional clock cycle to address generation.
We were able to install 4GB segments into the LDT, so any limits we imposed
on %cs and %ds were academic anyway. More importantly, this allows us to
make a page in the kernel readable to user applications, for holding things
like the signal trampoline and other fun things.
Move the user %cs/%ds segments from the LDT to the GDT. There was no good
reason for them to be there anyway. The old LDT entries are still there
but we can now relax the restriction that prevented users from emptying
the default LDT entries.
Putting user and kernel %cs and %ds together allows us to access the fast
sysenter/sysexit/syscall/sysret instructions. syscall/sysret in particular
require that the user/kernel segments be laid out this way. Reserve a slot
specifically for NDIS while here.
Create two user controllable slots in the GDT that are context switched
with the (kernel) thread. This allows user applications to set two
user privilige selectors to arbitary values. Create
i386_set_fsbase(void *base) and friends. (get/set, fs/gs). For i386,
%gs is used by tls and the thread libraries and this means that user
processes no longer have to have the cost of having a custom LDT, and
we will no longer to do a ldt switch when activating a kthread/ithread in
the usual case any more.
In other words, we can now set the base address for %fs and %gs to arbitary
addresses without the pain of messing with ldt segments.
sort and expand the prunelist in a more or less generally sensibly
order. Put some strategic comments in about how much storage is
necessary and a longer explanation on the top.
It is now pretty trivial to put nanoBSD on as little as a 64MB CF card.
peter [Wed, 13 Apr 2005 18:13:40 +0000 (18:13 +0000)]
Fix an evil bug that appeared in September 2003. VM86 bios calls use two
of the __pcb_spare longs. Except that fields were changed and one of the
spare values was used and the __pcb_spare field was reduced from two to one
long. Now VM86 bios calls can trash the first 4 bytes of the next page
following the kernel stack/pcb. This Is Bad(TM). This bug has been
present in 5.2-release and onwards, and is still in RELENG_5.
Instead of tempting fate and trying to use "spare" fields, explicitly
reserve them.
Use pci_find_bsf() to retrieve the PCI device associated with
a bus/device/function tuple.
This change enables pciconf(8) to work with CardBus devices.
Since cbb implements the pcib_ interface, it must also implement the
pcib_route_interrupt interface. Since there's only one interrupt pin
in the CardBus form factor, everybody gets to share it. Implement
cbb_route_interrupt to return the interrupt we have.
Default to a interrupt router that returns an invalid interrupt.
Otherwise, busses that implement the pcib interface that forget to
implement pcib_route_interrupt would return EIO, which the caller
interprets as 'use interrupt 6'. This is likely the cause of much of
the grief that we had when I enabled power modes for the cardbus
bridge, since the card needed to reroute the interrupt to it and it
was getting 6 which was d by the pccbb sanity checks.
Never hardcode /sys into these Makefiles. The proper way to spell it is $S.
Also, move the -I stuff to the centralized kern.pre.mk. However, it
might be better to add these flags to files.conf. This is a short
term fix to fix the broken builds on my machine (I don't have a valid
/sys link).
According to the comment in struct tty, t_modem is optional; hence we should
guard against NULL t_modem entry. Otherwise, driver doesn't have t_modem
callback implemented(such like sys/dev/usb/ucycom.c) would panic when
someone opens the driver's associated tty device.
harti [Wed, 13 Apr 2005 11:40:44 +0000 (11:40 +0000)]
Split VarFind() into a series of functions tailored for the different sets
of flags originally passed to VarFind(). This eliminates the code by
removing a bunch of tests.
jeff [Wed, 13 Apr 2005 10:57:13 +0000 (10:57 +0000)]
- Further simplify lookup; Force all filesystems to relock in the DOTDOT
case. There are bugs in some which didn't unlock in the ISDOTDOT case
to begin with that need to be addressed seperately. This simplifies
things anyway.
- Fix relookup() to prevent it from vrele()'ing the dvp while the vp
is locked. Catch up to other lookup changes.
Sponsored by: Isilon Systems, Inc.
Reported by: Peter Wemm
sos [Wed, 13 Apr 2005 07:14:17 +0000 (07:14 +0000)]
Add a ata_setmode method so we dont panic on setmode.
Note that the mode is only set on the device, we (mostly) have
no knowledge on how to set mode on the controller if at all possible.
Build cpufreq on ia64. The upcoming Montecito processor supports the
Enhanced SpeedStep (that is, a follow-up of it called Foxton). Until
we actually have support for that, we build to catch regressions in
the framework.
Implement unix(4) socket options LOCAL_CREDS and LOCAL_CONNWAIT.
- Add unp_addsockcred() (for LOCAL_CREDS).
- Add an argument to unp_connect2() to differentiate between
PRU_CONNECT and PRU_CONNECT2. (for LOCAL_CONNWAIT)
The latest release of the FreeBSD driver (twa) for
3ware's 9xxx series controllers. This corresponds to
the 9.2 release (for FreeBSD 5.2.1) on the 3ware website.
Highlights of this release are:
1. The driver has been re-architected to use a "Common Layer"
(all tw_cl* files), which is a consolidation of all OS-independent
parts of the driver. The FreeBSD OS specific portions of the
driver go into an "OS Layer" (all tw_osl* files).
This re-architecture is to achieve better maintainability, consistency
of behavior across OS's, and better portability to new OS's (drivers
for new OS's can be written by just adding an OS Layer that's specific
to the OS, by complying to a "Common Layer Programming Interface" API.
2. The driver takes advantage of multiple processors.
3. The driver has a new firmware image bundled, the new features of which
include Online Capacity Expansion and multi-lun support, among others.
More details about 3ware's 9.2 release can be found here:
http://www.3ware.com/download/Escalade9000Series/9.2/9.2_Release_Notes_Web.pdf
Since the Common Layer is used across OS's, the FreeBSD specific include
path for header files (/sys/dev/twa) is not part of the #include pre-processor
directive in any of the source files. For being able to integrate twa into
the kernel despite this, Makefile.<arch> has been changed to add the include
path to CFLAGS.
Dot the i's:
1 Move the debug.clock_adjust_* sysctls to debug.clock.adjust_* to
make it easier to get only the clock statistics.
2 Make the sysctls read-only [suggested by Marius].
3 When determining the new clock adjustment, we checked for an error
either larger than 12.5% or smaller than 12.5%. We left out an error
of exactly 12.5%. For errors larger than 12.5% we adjust the clock
reload value in such a way that the next clock interrupt would be
early (as in premature). For errors less than 12.5% we stopped the
adjustment.
The current algorithm doesn't benefit from excluding an error of
exactly 12.5%. Change the code to stop adjusting the clock if the
error is *not* larger than 12.5% [suggested by Marius].
sam [Tue, 12 Apr 2005 17:55:13 +0000 (17:55 +0000)]
Revise crypto api lightly to improve group key handling:
o don't pre-assign key index to the global key table entries so device
has a chance to decide what to use
o make ieee80211_crypto_newkey take the desired flags as an argument
instead of wacking the key structure directly; this eliminates a
bunch of code warts
o add a new flag IEEE80211_KEY_GROUP to indicate a key is a WPA Group
key so devices don't need to guess (temporarily add this flag in the
ioctl code until we can get wpa_supplicant+hostapd updated)
o shuffle IEEE80211_KEY_* bits to move flags used internally to the high
nibble of the flags word
Big cleanup of resource code for pccard. Once coventry noticed
problems here, it became clear we were being too complex.
o Don't keep track of resources in two places
o Use resource_list_purge instead of rolling our own
o Just reassign the ownership of the resource, rather than freeing it
and reallocating it.
o Fix compile problems when sizeof(u_long) != sizeof(int)
thomas [Tue, 12 Apr 2005 15:02:57 +0000 (15:02 +0000)]
(handler): When exitting upon an abnormal signal, yppush_exit() should not
attempt to clear its pending jobs list, as this could trigger another
signal, and cause an infinite recursion. What yppush_exit() tests
in order to determine whether to flush pending jobs is the
yppush_joblist chained list, so this is what needs to be cleared
in that case (not the yppush_jobs counter).
Reviewed by: audit@ (no objection)
MFC after: 2 weeks
Protect against recursive labels creation in simlar way as it is done
in BSD and MBR classes, ie. if provider below us uses the same metadata,
don't create labels based on the metadata.
This allows to create labels on geoms with rank != 1 without hacks.
Tested by: Chris Elsworth <chris@shagged.org> on sparc64
OK'ed by: phk
MFC after: 2 weeks
Cleanup of resource allocation code after having my attention focused on
this code:
o rid is stored in the resource, so don't bother keeping track of it here.
o Implement memory space
o Don't try to activate 'memory card' CFEs. This is type memory, as opposed
to the memory resource.
Use return value of resource_list_add to avoid a second
resource_list_find. Check to make sure that rle is not NULL and panic
if it is (but it appears that resource_list_add already panics, so I'm
not entirely sure it is necessary now).
Add a test to make sure we have a interrupt resource when we're
disabling it. This is also a cannot happen, but the extra care
shoudln't hurt.