]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoAdd an rc script for powerd(8).
njl [Sat, 26 Feb 2005 21:18:54 +0000 (21:18 +0000)]
Add an rc script for powerd(8).

19 years agoHook powerd up to the build.
njl [Sat, 26 Feb 2005 21:18:20 +0000 (21:18 +0000)]
Hook powerd up to the build.

19 years agoImport powerd(8), a utility for managing power consumption. Currently, it
njl [Sat, 26 Feb 2005 21:17:31 +0000 (21:17 +0000)]
Import powerd(8), a utility for managing power consumption.  Currently, it
just uses cpufreq(4) but in the future it should also do things like spin
down disks.  This is based on the work by Colin Percival (called "estctrl").

19 years agoMake power_profile not touch cpufreq by default.
njl [Sat, 26 Feb 2005 21:01:22 +0000 (21:01 +0000)]
Make power_profile not touch cpufreq by default.

19 years agoAdd the ability to specify "NONE" if the user wants no change for the
njl [Sat, 26 Feb 2005 20:17:07 +0000 (20:17 +0000)]
Add the ability to specify "NONE" if the user wants no change for the
given power profile.

MFC after: 1 day

19 years agoBring in a more healthy version of the libpthread for arm, which uses
cognet [Sat, 26 Feb 2005 19:06:49 +0000 (19:06 +0000)]
Bring in a more healthy version of the libpthread for arm, which uses
ARM_TP_ADDRESS.

19 years agoInstead of using sysarch() to store-retrieve the tp, add a magic address,
cognet [Sat, 26 Feb 2005 18:59:01 +0000 (18:59 +0000)]
Instead of using sysarch() to store-retrieve the tp, add a magic address,
ARM_TP_ADDRESS, where the tp will be stored. On CPUs that support it, a cache
line will be allocated and locked for this address, so that it will never go
to RAM. On CPUs that does not, a page is allocated for it (it will be a bit
slower, and is wrong for SMP, but should be fine for UP).
The tp is still stored in the mdthread struct, and at each context switch,
ARM_TP_ADDRESS gets updated.

Suggested by:   davidxu

19 years agoplug memory leak
sam [Sat, 26 Feb 2005 18:55:53 +0000 (18:55 +0000)]
plug memory leak

Noticed by: Coverity Prevent analysis tool

19 years agoMinor cleanup in nfs_request() and removal of a comment that doesn't
ps [Sat, 26 Feb 2005 18:55:36 +0000 (18:55 +0000)]
Minor cleanup in nfs_request() and removal of a comment that doesn't
reflect reality.

Submitted by: Mohan Srinivasan

19 years agomove ptr deref's to after null checks
sam [Sat, 26 Feb 2005 18:52:54 +0000 (18:52 +0000)]
move ptr deref's to after null checks

Noticed by: Coverity Prevent analysis tool

19 years agoNew release notes: moused(8) -V (+MFC).
bmah [Sat, 26 Feb 2005 16:33:04 +0000 (16:33 +0000)]
New release notes:  moused(8) -V (+MFC).

MFCs noted:  TCP random ephemeral port rate limiting, syslogd(8)
logpriv (with a few grammar fixes), tail -f,

Other changes:  Minor grammar fixes to rc.conf interface renaming and
rc.d/moused notes.

19 years agoRemove carp_softc.sc_ifp member in favor of union pointers in struct ifnet.
glebius [Sat, 26 Feb 2005 13:55:07 +0000 (13:55 +0000)]
Remove carp_softc.sc_ifp member in favor of union pointers in struct ifnet.

Obtained from: OpenBSD

19 years agoAdd a neat remark on the popular belief that the universe is a safe
yar [Sat, 26 Feb 2005 12:49:03 +0000 (12:49 +0000)]
Add a neat remark on the popular belief that the universe is a safe
the man ought to break open and rob.

19 years agoMark _rtld_allocate_tls and _rtld_free_tls as weak symbols for Variant I
davidxu [Sat, 26 Feb 2005 10:39:49 +0000 (10:39 +0000)]
Mark _rtld_allocate_tls and _rtld_free_tls as weak symbols for Variant I
tls.

Reviewed by: dfr

19 years agoStaticize local functions.
glebius [Sat, 26 Feb 2005 10:33:14 +0000 (10:33 +0000)]
Staticize local functions.

19 years agoUn-document the non-extant exp10() and exp10f() functions.
das [Sat, 26 Feb 2005 08:54:45 +0000 (08:54 +0000)]
Un-document the non-extant exp10() and exp10f() functions.
exp10() was a casualty of the transition away from the VAX.

19 years agoMajor clean up and split the Var_Parse monster function into three
harti [Sat, 26 Feb 2005 08:54:40 +0000 (08:54 +0000)]
Major clean up and split the Var_Parse monster function into three
functions: one for the single letter variables, one for the others
and one that does the recursive expansion.

Patches: 7.68-7.79

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoDon't need to do MmInitializeMdl() in ndis_mtop() anymore:
wpaul [Sat, 26 Feb 2005 07:11:17 +0000 (07:11 +0000)]
Don't need to do MmInitializeMdl() in ndis_mtop() anymore:
IoInitializeMdl() does it internally (and doing it again here
messes things up).

19 years agoMFCs noted: cpufreq(4) (partial), matcd(4) removed, rm(1) -I.
bmah [Sat, 26 Feb 2005 05:24:40 +0000 (05:24 +0000)]
MFCs noted:  cpufreq(4) (partial), matcd(4) removed, rm(1) -I.

Other changes:  Minor punctuation fixes in hptmv(4) note.

19 years agoFix mismerge.
trhodes [Sat, 26 Feb 2005 04:14:20 +0000 (04:14 +0000)]
Fix mismerge.

Noticed by: tjr

19 years agomove ptr use down to after null check
sam [Sat, 26 Feb 2005 02:31:34 +0000 (02:31 +0000)]
move ptr use down to after null check

Noticed by: Coverity Prevent analysis tool
Reviewed by: emax

19 years agostyle(9).
obrien [Sat, 26 Feb 2005 01:19:21 +0000 (01:19 +0000)]
style(9).

19 years agoDeclare the sbus(4) front-end of puc(4) also for fhc(4), allowing
marius [Sat, 26 Feb 2005 00:25:43 +0000 (00:25 +0000)]
Declare the sbus(4) front-end of puc(4) also for fhc(4), allowing
uart(4) to support the Zilog 8530 SCCs which hang off of a FireHose
bus on Sun E4000/E5000 class machines.
Beside the fact that a puc_fhc.c would just be a copy of puc_sbus.c
with s,sbus,fhc,g the reason why the declaration for fhc(4) was
sticked into puc_sbus.c is that both of these front-ends for puc(4)
will go away once there is a scc(4).

Discussed with: marcel
Tested by: hrs, kris
MFC after: 3 days

19 years agoAdd a stopgap allowing puc(4) to allocate IRQs on fhc(4). Given that
marius [Sat, 26 Feb 2005 00:22:52 +0000 (00:22 +0000)]
Add a stopgap allowing puc(4) to allocate IRQs on fhc(4). Given that
both a scc(4) is under way and fhc(4) will be change to use INOs this
shouldn't stay in HEAD for too long but we need a MFC-able solution
for FreeBSD 5.4.

Discussed with: marcel
Tested by: hrs, kris
MFC after: 3 days

19 years agoMDLs are supposed to be variable size (they include an array of pages
wpaul [Sat, 26 Feb 2005 00:22:16 +0000 (00:22 +0000)]
MDLs are supposed to be variable size (they include an array of pages
that describe a buffer of variable size). The problem is, allocating
MDLs off the heap is slow, and it can happen that drivers will allocate
lots and lots of lots of MDLs as they run.

As a compromise, we now do the following: we pre-allocate a zone for
MDLs big enough to describe any buffer with 16 or less pages. If
IoAllocateMdl() needs a MDL for a buffer with 16 or less pages, we'll
allocate it from the zone. Otherwise, we allocate it from the heap.
MDLs allocate from the zone have a flag set in their mdl_flags field.
When the MDL is released, IoMdlFree() will uma_zfree() the MDL if
it has the MDL_ZONE_ALLOCED flag set, otherwise it will release it
to the heap.

The assumption is that 16 pages is a "big number" and we will rarely
need MDLs larger than that.

- Moved the ndis_buffer zone to subr_ntoskrnl.c from kern_ndis.c
  and named it mdl_zone.

- Modified IoAllocateMdl() and IoFreeMdl() to use uma_zalloc() and
  uma_zfree() if necessary.

- Made ndis_mtop() use IoAllocateMdl() instead of calling uma_zalloc()
  directly.

Inspired by: discussion with Giridhar Pemmasani

19 years agoRemove duplicate FBSDID.
marius [Sat, 26 Feb 2005 00:21:25 +0000 (00:21 +0000)]
Remove duplicate FBSDID.

19 years agoAdd the field in the md part of the struct thread required by ARM_[GET|SET]_TP.
cognet [Sat, 26 Feb 2005 00:02:14 +0000 (00:02 +0000)]
Add the field in the md part of the struct thread required by ARM_[GET|SET]_TP.

19 years agokill unused variable
sam [Fri, 25 Feb 2005 23:15:48 +0000 (23:15 +0000)]
kill unused variable

Noticed by: Coverity Prevent analysis tool

19 years agoQuiet error messages if the requested sysctls are not present.
njl [Fri, 25 Feb 2005 23:14:41 +0000 (23:14 +0000)]
Quiet error messages if the requested sysctls are not present.

MFC after: 1 day

19 years agokill unused value
sam [Fri, 25 Feb 2005 23:06:25 +0000 (23:06 +0000)]
kill unused value

Noticed by: Coverity Prevent analysis tool

19 years agoremove dead code
sam [Fri, 25 Feb 2005 22:58:25 +0000 (22:58 +0000)]
remove dead code

Noticed by: Coverity Prevent analysis tool

19 years agoImplement two new sysarch for arm, ARM_GET_TP and ARM_SET_TP, to work around
cognet [Fri, 25 Feb 2005 22:56:16 +0000 (22:56 +0000)]
Implement two new sysarch for arm, ARM_GET_TP and ARM_SET_TP, to work around
the lack of tls on arm.

19 years agoMake sure casuptr() reset pcb->pcb_onfault when returning.
cognet [Fri, 25 Feb 2005 22:54:56 +0000 (22:54 +0000)]
Make sure casuptr() reset pcb->pcb_onfault when returning.

19 years agoRemove recently added note about DEVICE_POLLING not working with SMP.
trhodes [Fri, 25 Feb 2005 22:07:51 +0000 (22:07 +0000)]
Remove recently added note about DEVICE_POLLING not working with SMP.
Remove warning from kern_poll.c to allow DEVICE_POLLING to be built with SMP.

Discussed with: ru, glebius

19 years agoRemove unused variable.
njl [Fri, 25 Feb 2005 22:01:58 +0000 (22:01 +0000)]
Remove unused variable.

Noticed by: Coverity tool

19 years agoavoid null ptr deref
sam [Fri, 25 Feb 2005 21:57:46 +0000 (21:57 +0000)]
avoid null ptr deref

Noticed by: Coverity Prevent analysis tool
Reviewed by: scottl

19 years agoInsert missing increment of (i) when walking the temporary semaphore
rwatson [Fri, 25 Feb 2005 21:00:14 +0000 (21:00 +0000)]
Insert missing increment of (i) when walking the temporary semaphore
vector during fork.

Fix assertion which contained an off-by-one error.

Submitted by: Antoine Brodin < antoine dot brodin at laposte dot net >

19 years agoo avoid potential null ptr deref if symbol lookup fails
sam [Fri, 25 Feb 2005 20:50:20 +0000 (20:50 +0000)]
o avoid potential null ptr deref if symbol lookup fails
o unload module if symbol lookup fails

Noticed by: Coverity Prevent analysis tool (null ptr deref)
Reviewed by: bms, imp, dwhite

19 years agocorrect bounds check of fd parameter
sam [Fri, 25 Feb 2005 20:15:11 +0000 (20:15 +0000)]
correct bounds check of fd parameter

Noticed by: Coverity Prevent analysis tool

19 years agoAdd notes on merging acpi_perf and acpi_throttle to acpi.ko
njl [Fri, 25 Feb 2005 20:10:38 +0000 (20:10 +0000)]
Add notes on merging acpi_perf and acpi_throttle to acpi.ko

19 years agofail gracefully rather than using an invalid array index if unable
sam [Fri, 25 Feb 2005 19:47:18 +0000 (19:47 +0000)]
fail gracefully rather than using an invalid array index if unable
to allocate a bar; it's unclear whether this can happen in practice

Noticed by: Coverity Prevent analysis tool
Discussed with: marcel

19 years agoChange the definition of struct if_data's member ifi_epoch from wall
brooks [Fri, 25 Feb 2005 19:46:41 +0000 (19:46 +0000)]
Change the definition of struct if_data's member ifi_epoch from wall
clock time to uptime because wall clock time may go backwards.

This is a change in the API which will impact SNMP agents who are using
ifi_epoch to set RFC2233's ifCounterDiscontinuityTime.  None are know to
exist today.  This will not impact applications that are using the
<index, epoch> tuple to verify interface uniqueness except that it
eliminates a race which could lead to a false assumption of uniqueness.

Because this is a behavior change, bump __FreeBSD_version.

Discussed with: re (jhb, scottl)
MFC after: 3 days
Pointed out by: pkh (way back at EuroBSDCon)
Pointy hat: brooks

19 years agofixup signal mapping:
sam [Fri, 25 Feb 2005 19:34:10 +0000 (19:34 +0000)]
fixup signal mapping:
o change the mapping arrays to have a zero offset rather than base 1;
  this eliminates lots of signo adjustments and brings the code
  back inline with the original netbsd code
o purge use of SVR4_SIGTBLZ; SVR4_NSIG is the only definition for
  how big a mapping array is
o change the mapping loops to explicitly ignore signal 0
o purge some bogus code from bsd_to_svr4_sigset
o adjust svr4_sysentvec to deal with the mapping table change

Enticed into fixing by: Coverity Prevent analysis tool
Glanced at by: marcel, jhb

19 years agoResort the CPU list with regards to recently added CPUs.
trhodes [Fri, 25 Feb 2005 19:32:51 +0000 (19:32 +0000)]
Resort the CPU list with regards to recently added CPUs.

Prodded by: glewis

19 years agoAdd an exit hook, sem_forkhook(), which walks the list of POSIX semaphores
rwatson [Fri, 25 Feb 2005 19:10:51 +0000 (19:10 +0000)]
Add an exit hook, sem_forkhook(), which walks the list of POSIX semaphores
owned by a process when it forks, and creates a matching set of references
for the child process, as prescribed by POSIX.

In order to avoid races with other threads in the parent process during
fork(), it is necessary to allocate a temporary reference list while
holding the sem_lock, then transfer those references to the new process
once the sem_lock is released.  The implementation is inefficient but
appears functional; in order to improve the efficiency, it will be
necessary to modify the existing structures and logic, which generally
rely on O(n) operations over the global set of semaphores.

19 years agoAdd macros to construct Windows IOCTL codes, and to extract function
wpaul [Fri, 25 Feb 2005 18:25:48 +0000 (18:25 +0000)]
Add macros to construct Windows IOCTL codes, and to extract function
codes from an IOCTL. (The USB module will need them later.)

19 years agoAdd a missing comma.
pjd [Fri, 25 Feb 2005 17:22:23 +0000 (17:22 +0000)]
Add a missing comma.

19 years agoAssert sem_lock in id_to_sem() and sem_lookup_byname(), since these
rwatson [Fri, 25 Feb 2005 17:01:35 +0000 (17:01 +0000)]
Assert sem_lock in id_to_sem() and sem_lookup_byname(), since these
functions iterate over the global POSIX semaphore lists.

MFC after: 3 days

19 years agoInstead of assuming units of bytes, it seems more likely that this is
njl [Fri, 25 Feb 2005 16:57:34 +0000 (16:57 +0000)]
Instead of assuming units of bytes, it seems more likely that this is
a bitfield.

19 years agoNew release notes:
hrs [Fri, 25 Feb 2005 16:40:37 +0000 (16:40 +0000)]
New release notes:
syslogd(8) /var/run/logpriv,
tail -f multiple file support, and
ehci(4) improved.

Update release notes:
sort "rm -I" entry,
add the reasons why T/TCP support removed, and
remove duplicated snd_audiocs(4) entry[1].

Pointed out by: bmah[1]

19 years agoFix prototype for bpf_mtap2().
glebius [Fri, 25 Feb 2005 15:58:05 +0000 (15:58 +0000)]
Fix prototype for bpf_mtap2().

MFC after: 3 days

19 years agoNew release notes: HZ change, MAXSHELLCMDLEN change, Heimdal 0.6.3,
bmah [Fri, 25 Feb 2005 15:46:10 +0000 (15:46 +0000)]
New release notes:  HZ change, MAXSHELLCMDLEN change, Heimdal 0.6.3,
OpenSSL 0.9.7e.

MFC noted:  libarchive/bsdtar ISO/ZIP support.

19 years agoChange the return value of Var_Subst to return a Buffer instead
harti [Fri, 25 Feb 2005 13:16:56 +0000 (13:16 +0000)]
Change the return value of Var_Subst to return a Buffer instead
of a char *.

Patch: 7.49

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoWelcome to the 21st century: increase MAXSHELLCMDLEN from 128 bytes to
sobomax [Fri, 25 Feb 2005 11:49:42 +0000 (11:49 +0000)]
Welcome to the 21st century: increase MAXSHELLCMDLEN from 128 bytes to
PAGE_SIZE.

Unlike originator of the PR suggests retain MAXSHELLCMDLEN definition
(he has been proposing to replace it with PAGE_SIZE everywhere), not only
this reduced the diff significantly, but prevents code obfuscation and also
allows to increase/decrease this parameter easily if needed.

PR: kern/64196
Submitted by: Magnus Bäckström <b@etek.chalmers.se>

19 years agoNew lines when logging.
glebius [Fri, 25 Feb 2005 11:26:39 +0000 (11:26 +0000)]
New lines when logging.

19 years agoEmbrace macros with do {} while (0)
glebius [Fri, 25 Feb 2005 10:49:47 +0000 (10:49 +0000)]
Embrace macros with do {} while (0)

Submitted by: maxim

19 years agoUpdate description of net.inet.carp.log
glebius [Fri, 25 Feb 2005 10:38:43 +0000 (10:38 +0000)]
Update description of net.inet.carp.log

19 years agoo Replace two while {} do loops with more appropriate do {} while loops. This
sobomax [Fri, 25 Feb 2005 10:17:53 +0000 (10:17 +0000)]
o Replace two while {} do loops with more appropriate do {} while loops. This
  doesn't change functionality, but makes code more logical.

Obtained from: DrafonFlyBSD

o Use VOP_GETATTR() to obtain actual size of file and parse no more than that.
  Previously, we parsed MAXSHELLCMDLEN characters regardless of the actual file
  size. This makes the following working:

$ printf '#!/bin/echo' > /tmp/test.sh
$ chmod 755 /tmp/test.sh
$ /tmp/test.sh

Previously, attempts to execve() that shell script has been failing with bogus
ENAMETOOLONG.

PR: kern/64196
Submitted by: Magnus B.ckstr.m <b@etek.chalmers.se>

19 years agoCall carp_carpdev_state() from carp_set_addr6(). See log for rev 1.4.
glebius [Fri, 25 Feb 2005 10:12:11 +0000 (10:12 +0000)]
Call carp_carpdev_state() from carp_set_addr6(). See log for rev 1.4.

Sponsored by: Rambler

19 years agoImprove logging:
glebius [Fri, 25 Feb 2005 10:09:44 +0000 (10:09 +0000)]
Improve logging:
 - Simplify CARP_LOG() and making it working (we don't have addlog in FreeBSD).
 - Introduce CARP_DEBUG() which logs with LOG_DEBUG severity when
   net.inet.carp.log > 1
 - Use CARP_DEBUG to log state changes of carp interfaces.

After CARP_LOG() cleanup it appeared that carp_input_c() does not need sc
argument. Remove it.

Sponsored by: Rambler

19 years agoThe chip specific functions have been split out in their own
harti [Fri, 25 Feb 2005 09:49:29 +0000 (09:49 +0000)]
The chip specific functions have been split out in their own
C files to simplify adding of new PHY chips.
Include the split out .c files in the module build too.

19 years agoTry harder to not exceed MAXSHELLCMDLEN when parsing first line of shell
sobomax [Fri, 25 Feb 2005 08:42:04 +0000 (08:42 +0000)]
Try harder to not exceed MAXSHELLCMDLEN when parsing first line of shell
script. Otherwise it's possible to panic kernel by constructing a shell
script with first line not ending in '\n'.

Also, treat '\0' as line terminating character, which may me useful in
some situations.

Submitted by: gad

19 years agoFix a couple of callback instances that should have been wrapped with
wpaul [Fri, 25 Feb 2005 08:34:32 +0000 (08:34 +0000)]
Fix a couple of callback instances that should have been wrapped with
MSCALLx().

Add definition for STATUS_PENDING error code.

19 years agoRemove acpi_perf from {ARCH}/conf/NOTES, to make tinderbox happy.
delphij [Fri, 25 Feb 2005 07:10:37 +0000 (07:10 +0000)]
Remove acpi_perf from {ARCH}/conf/NOTES, to make tinderbox happy.

Reported by: tinderbox
Inspired by: acpi_perf build structure removal commit

19 years agoAdd missing ofw_bus_if.h to SRCS.
yongari [Fri, 25 Feb 2005 06:59:56 +0000 (06:59 +0000)]
Add missing ofw_bus_if.h to SRCS.

Submitted by: joerg

19 years agoCompute the right length to use with bzero() when initializing an IRP
wpaul [Fri, 25 Feb 2005 06:31:45 +0000 (06:31 +0000)]
Compute the right length to use with bzero() when initializing an IRP
in IoInitializeIrp() (must use IoSizeOfIrp() to account for the stack
locations).

19 years agoFile removed in update from OpenSSL 0.9.7d -> 0.9.7e.
nectar [Fri, 25 Feb 2005 06:22:30 +0000 (06:22 +0000)]
File removed in update from OpenSSL 0.9.7d -> 0.9.7e.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r142430,
nectar [Fri, 25 Feb 2005 06:14:53 +0000 (06:14 +0000)]
This commit was generated by cvs2svn to compensate for changes in r142430,
which included commits to RCS files with non-trunk default branches.

19 years agoUpdate OpenSSL 0.9.7d -> 0.9.7e.
nectar [Fri, 25 Feb 2005 06:04:12 +0000 (06:04 +0000)]
Update OpenSSL 0.9.7d -> 0.9.7e.

19 years agoResolve conflicts after import of OpenSSL 0.9.7e.
nectar [Fri, 25 Feb 2005 05:49:44 +0000 (05:49 +0000)]
Resolve conflicts after import of OpenSSL 0.9.7e.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r142425,
nectar [Fri, 25 Feb 2005 05:39:05 +0000 (05:39 +0000)]
This commit was generated by cvs2svn to compensate for changes in r142425,
which included commits to RCS files with non-trunk default branches.

19 years agoVendor import of OpenSSL 0.9.7e.
nectar [Fri, 25 Feb 2005 05:39:05 +0000 (05:39 +0000)]
Vendor import of OpenSSL 0.9.7e.

19 years agoGet SYSDIR set correctly for building ports.
imp [Fri, 25 Feb 2005 05:34:45 +0000 (05:34 +0000)]
Get SYSDIR set correctly for building ports.
On install, do deinstall reinstall

19 years agoUpdate list of files to remove prior to import of OpenSSL 0.9.7e.
nectar [Fri, 25 Feb 2005 05:31:23 +0000 (05:31 +0000)]
Update list of files to remove prior to import of OpenSSL 0.9.7e.

19 years agoClean up the OpenSSL vendor branch by removing files that are not
nectar [Fri, 25 Feb 2005 05:25:37 +0000 (05:25 +0000)]
Clean up the OpenSSL vendor branch by removing files that are not
part of recent releases.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r142421,
nectar [Fri, 25 Feb 2005 05:25:37 +0000 (05:25 +0000)]
This commit was generated by cvs2svn to compensate for changes in r142421,
which included commits to RCS files with non-trunk default branches.

19 years agoAdd back 'n' that went missing in last commit
imp [Fri, 25 Feb 2005 03:43:43 +0000 (03:43 +0000)]
Add back 'n' that went missing in last commit

19 years agoplug memory leak
sam [Fri, 25 Feb 2005 03:41:11 +0000 (03:41 +0000)]
plug memory leak

Noticed by: Coverity Prevent analysis tool
Reviewed by: emax

19 years agoAdd PVO_FAKE flag to pvo entries for PG_FICTITIOUS mappings, to
grehan [Fri, 25 Feb 2005 02:42:15 +0000 (02:42 +0000)]
Add PVO_FAKE flag to pvo entries for PG_FICTITIOUS mappings, to
avoid trying to reverse-map a device physical address to the
vm_page array and walking into non-existent vm weeds.

found by:  Xorg server exiting

19 years agoCorrect e-mail address in copyright.
wpaul [Fri, 25 Feb 2005 02:36:23 +0000 (02:36 +0000)]
Correct e-mail address in copyright.

19 years agoMods for Xorg server:
grehan [Fri, 25 Feb 2005 02:29:01 +0000 (02:29 +0000)]
Mods for Xorg server:

 - store assigned PCI addresses at cninit time for later mmap range
   check
 - implement set_border to scrub X remnants when switching back to VTYs
 - implement mmap, only allowing addresses within the range of the
   console adapter.

19 years agouse __target in preference to target
imp [Fri, 25 Feb 2005 02:25:38 +0000 (02:25 +0000)]
use __target in preference to target

19 years agoFix a few markup nits in previous commit.
trhodes [Fri, 25 Feb 2005 00:40:46 +0000 (00:40 +0000)]
Fix a few markup nits in previous commit.

Noticed by: ru, who else? :)

19 years agoReword previous commit to be a bit more correct and provide more information.
trhodes [Fri, 25 Feb 2005 00:32:41 +0000 (00:32 +0000)]
Reword previous commit to be a bit more correct and provide more information.

Inspiried by: ru

19 years agoHandle endianness for arm.
cognet [Fri, 25 Feb 2005 00:24:03 +0000 (00:24 +0000)]
Handle endianness for arm.

19 years agoavoid sneaky double free
sam [Fri, 25 Feb 2005 00:16:53 +0000 (00:16 +0000)]
avoid sneaky double free

Noticed by: Coverity Prevent analysis tool
Reviewed by: scottl

19 years agoApparently, the probe routine in if_ndis_usb.c can be called twice
wpaul [Thu, 24 Feb 2005 22:54:15 +0000 (22:54 +0000)]
Apparently, the probe routine in if_ndis_usb.c can be called twice
for a given device in some circumstances, so move the PDO creation
to the attach routine so we don't end up creating two PDOs.

Also, when we skip the call to ndis_convert_res() in if_ndis.c:ndis_attach(),
initialize sc->ndis_block->nmb_rlist to NULL. We don't explicitly zero
the miniport block, so this will make sure ndis_unload_driver() does
the right thing.

19 years agoFix style(9) issues with __P removal.
imp [Thu, 24 Feb 2005 22:33:05 +0000 (22:33 +0000)]
Fix style(9) issues with __P removal.

Noticed by: bde

19 years agoUpdate Heimdal 0.6.1 -> 0.6.3.
nectar [Thu, 24 Feb 2005 22:24:24 +0000 (22:24 +0000)]
Update Heimdal 0.6.1 -> 0.6.3.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r142403,
nectar [Thu, 24 Feb 2005 22:22:53 +0000 (22:22 +0000)]
This commit was generated by cvs2svn to compensate for changes in r142403,
which included commits to RCS files with non-trunk default branches.

19 years agoVendor import of Heimdal 0.6.3.
nectar [Thu, 24 Feb 2005 22:22:53 +0000 (22:22 +0000)]
Vendor import of Heimdal 0.6.3.

19 years agoDo not include lib/kdfs in future imports.
nectar [Thu, 24 Feb 2005 22:17:13 +0000 (22:17 +0000)]
Do not include lib/kdfs in future imports.

19 years agoRemove lib/kdfs from vendor branch: we do not build it, and it will not
nectar [Thu, 24 Feb 2005 22:16:35 +0000 (22:16 +0000)]
Remove lib/kdfs from vendor branch: we do not build it, and it will not
be included in future imports.

19 years agoClean up the Heimdal vendor branch by removing files not included in
nectar [Thu, 24 Feb 2005 22:14:04 +0000 (22:14 +0000)]
Clean up the Heimdal vendor branch by removing files not included in
any import for several years.

If memory serves, this was
Suggested by: ru
an awfully long time ago-- sorry for the delay!

19 years ago- Correct one aspect of the driver_object/device_object/IRP framework:
wpaul [Thu, 24 Feb 2005 21:49:14 +0000 (21:49 +0000)]
- Correct one aspect of the driver_object/device_object/IRP framework:
  when we create a PDO, the driver_object associated with it is that
  of the parent driver, not the driver we're trying to attach. For
  example, if we attach a PCI device, the PDO we pass to the NdisAddDevice()
  function should contain a pointer to fake_pci_driver, not to the NDIS
  driver itself. For PCI or PCMCIA devices this doesn't matter because
  the child never needs to talk to the parent bus driver, but for USB,
  the child needs to be able to send IRPs to the parent USB bus driver, and
  for that to work the parent USB bus driver has to be hung off the PDO.

  This involves modifying windrv_lookup() so that we can search for
  bus drivers by name, if necessary. Our fake bus drivers attach themselves
  as "PCI Bus," "PCCARD Bus" and "USB Bus," so we can search for them
  using those names.

  The individual attachment stubs now create and attach PDOs to the
  parent bus drivers instead of hanging them off the NDIS driver's
  object, and in if_ndis.c, we now search for the correct driver
  object depending on the bus type, and use that to find the correct PDO.

  With this fix, I can get my sample USB ethernet driver to deliver
  an IRP to my fake parent USB bus driver's dispatch routines.

- Add stub modules for USB support: subr_usbd.c, usbd_var.h and
  if_ndis_usb.c. The subr_usbd.c module is hooked up the build
  but currently doesn't do very much. It provides the stub USB
  parent driver object and a dispatch routine for
  IRM_MJ_INTERNAL_DEVICE_CONTROL. The only exported function at
  the moment is USBD_GetUSBDIVersion(). The if_ndis_usb.c stub
  compiles, but is not hooked up to the build yet. I'm putting
  these here so I can keep them under source code control as I
  flesh them out.

19 years agoReturn BUS_PROBE_DEFAULT instead of 0.
imp [Thu, 24 Feb 2005 21:32:56 +0000 (21:32 +0000)]
Return BUS_PROBE_DEFAULT instead of 0.

19 years agoMove acpi_perf and acpi_throttle into acpi.ko. Remove the acpi_perf
njl [Thu, 24 Feb 2005 20:48:07 +0000 (20:48 +0000)]
Move acpi_perf and acpi_throttle into acpi.ko.  Remove the acpi_perf
build structure.

19 years agoNote the ACPI clock and its characteristics.
njl [Thu, 24 Feb 2005 20:28:41 +0000 (20:28 +0000)]
Note the ACPI clock and its characteristics.

19 years agoBump the maximum number of levels to 64 and add warning messages about
njl [Thu, 24 Feb 2005 20:21:41 +0000 (20:21 +0000)]
Bump the maximum number of levels to 64 and add warning messages about
what to do to fix reduced functionality if the number of levels is too low.

19 years agoCorrect an off-by-one error in the number of settings est announces.
njl [Thu, 24 Feb 2005 20:20:11 +0000 (20:20 +0000)]
Correct an off-by-one error in the number of settings est announces.
The extraneous "0" state was not fatal but useless.