Commit graph

4955 commits

Author SHA1 Message Date
Arun Thomas
32aabfab40 ramdisk: build pwd_mkdb 2011-11-22 16:45:17 +01:00
Ben Gras
cd107a5ac3 fxp: add device id in the right place 2011-11-22 11:17:49 +01:00
Ben Gras
81bdb22b0c fxp: add device id for LU82551IT
Contributed by Roman Ignatov
2011-11-22 11:09:42 +01:00
David van Moolenbroek
f29ba928cc isofs: unbreak lseek 2011-11-22 11:46:55 +01:00
Ben Gras
8b69d04fff csu: add code to call constructors to be used by clang
. clang-linked binaries were not calling global constructors, as the
	  code to do so wasn't in csu/ and linked
	. it does work for gcc as it uses its self-supplied crt{begin,end} code
	. this commit copies netbsd's crt{begin,end}.S, which contains
	  constructor/destructor calling code, called from .init and .fini
	  sections already accumulated by the linker. the _init function was already
	  called by the C startup code before calling main.
	. based on work by Antoine Leca
2011-11-20 20:20:51 +01:00
Thomas Cort
0c9b539a35 compat.S: add getpwnam() alias.
Needed for finding getpwnam() when a prototype is not given.
2011-11-20 18:36:42 +01:00
Ben Gras
c24d15b2db pm: add mproc table sanity check feature
. make procfs check it
	. detects pm/procfs mismatches
	. was triggered by ack/clang pm/procfs:
	  add padding to mproc struct to align ack/clang layout
	  to fix this
2011-11-18 17:18:10 +01:00
Ben Gras
b1eba81b9d simplify build logic a bit
. always compile acpi, with clang, so never have
	  build/clean inconsistencies; can be enabled (i.e. run
	  at boot time) by setting acpi variable in the boot monitor
	. always strip binaries with the right strip cmd, so never
	  have ack/elf strip inconsistencies
2011-11-18 17:11:17 +01:00
Arun Thomas
3f9ef4ac46 make: set MACHINE_ARCH variable
Also, modify sys.mk so we continue to build with -O
2011-11-18 16:22:25 +01:00
Ben Gras
f379b3eecb remove ftpd200 2011-11-18 12:19:50 +01:00
Ben Gras
ee8602a486 minor bugfixes caught by clang warnings 2011-11-18 12:18:23 +01:00
Ben Gras
79285c3664 libarchive: not for ack
. fixes compile error due to missing bzlib.h for ack
	. one less ack library
2011-11-18 11:49:21 +01:00
Vivek Prakash
79bfef9aab import NetBSD bzip2 and libbz2 2011-11-18 11:11:04 +01:00
Ben Gras
349a158056 base system hooks for pkgsrc drivers
. rc script and service know to look in /usr/pkg/.. for
	  extra binaries and conf files
	. service split into parsing config and doing RS request
	  so that a new utility (printconfig) can just print the
	  config in machine-parseable format for netconf integration
	. converted all base system eth drivers/netconf
2011-11-16 16:18:21 +01:00
Ben Gras
2c57ee566b pwcache_userdb alias.
. mainly so configure can find it without the prototype.
2011-11-16 12:59:40 +01:00
David van Moolenbroek
34c14b1cab More cleanup: no more CMOS and SCSI ioctls 2011-11-16 11:50:53 +01:00
Vivek Prakash
c4bce95897 Define flags in sys/stat.h 2011-11-15 17:14:57 +01:00
Ben Gras
41f823ae4c release.sh: add -e
. allows extra files in release for flexibility
	. written by Jan Wieck
2011-11-15 14:18:56 +01:00
Ben Gras
0811b47a71 df: add case for MFSv3
. fix for FSes mounted with '-t mfs' syntax
2011-11-14 22:18:31 +00:00
Ben Gras
5c00743626 Switch to NetBSD passwd format
Based on work by Vivek Prakash and Gianluca Guida.

See UPDATING about caveats on currently existing accounts.

	. restores netbsd libc pwcache functions
2011-11-14 14:47:42 +00:00
Ben Gras
cedeabb357 mainstream-format fstab format
. detect both formats in /etc/rc
	. generate new format in setup
	. obsoletes /etc/fstab.local: everything can go in /etc/fstab
	. put shutdown/reboot/halt and a copy of /usr/adm/wtmp
	  (/etc/wtmp) on root FS so that we can do shutdown checks before
	  mounting /usr
	. new fstab format makes getfsent() and friends work
2011-11-14 14:47:28 +00:00
Thomas Veerman
4311136125 Fix compiler warnings 2011-11-14 14:21:39 +00:00
Thomas Veerman
490e0de548 Import librefuse and libpuffs
Import libpuffs and our port of libpuffs. The port was done as part of
GSoC 2011 FUSE project, done by Evgeniy Ivanov. The librefuse import
did not require any porting efforts. Libpuffs has been modified to
understand our VFS-FS protocol and translate between that and PUFFS. As
an example that it works, fuse-ntfs-3g from pkgsrc can be compiled and
used to mount ntfs partitions:
mount -t ntfs-3g <device> <mountpoint>

FUSE only works with the asynchronous version of VFS. See <docs/UPDATING> on
how to run AVFS.

This patch further includes some changes to mount(1) and mount(2) so it's
possible to use file systems provided by pkgsrc (note: manual modifications
to /etc/system.conf are still needed. There has been made an exception for
fuse-ntfs-3g, so it already as an entry).
2011-11-14 11:53:05 +00:00
Thomas Veerman
a209c3ae12 Fix a ton of compiler warnings
This patch fixes most of current reasons to generate compiler warnings.
The changes consist of:
 - adding missing casts
 - hiding or unhiding function declarations
 - including headers where missing
 - add __UNCONST when assigning a const char * to a char *
 - adding missing return statements
 - changing some types from unsigned to signed, as the code seems to want
   signed ints
 - converting old-style function definitions to current style (i.e.,
   void func(param1, param2) short param1, param2; {...} to
   void func (short param1, short param2) {...})
 - making the compiler silent about signed vs unsigned comparisons. We
   have too many of those in the new libc to fix.

A number of bugs in the test set were fixed. These bugs were never
triggered with our old libc. Consequently, these tests are now forced to
link with the new libc or they will generate errors (in particular tests 43
and 55).

Most changes in NetBSD libc are limited to moving aroudn "#ifndef __minix"
or stuff related to Minix-specific things (code in sys-minix or gen/minix).
2011-11-14 10:07:49 +00:00
David van Moolenbroek
85b8fbe72b More cleanup of minix/com.h. 2011-11-11 13:53:29 +01:00
David van Moolenbroek
b477504754 ps(1): remove outdated comment 2011-11-11 11:37:44 +01:00
David van Moolenbroek
7207cf2e37 ps(1) man page: mem/kmem are no longer used 2011-11-11 11:34:50 +01:00
David van Moolenbroek
b1855cdcf2 Change default pager to less(1)
For one thing, less(1) provides better interaction with git.
2011-11-10 16:58:46 +01:00
David van Moolenbroek
af01bda509 libbdev: initial version
The "bdev" library provides basic primitives for file systems to talk
to block device drivers, hiding the details of the underlying protocol
and interaction model.

This version of libbdev is rather basic. It is planned to support the
following features in the long run:

 - asynchronous requests and replies;
 - recovery support for underlying block drivers;
 - retrying of failed I/O requests.

The commit also changes our block-based file systems (mfs, ext2, isofs)
to make use of libbdev.
2011-11-09 14:43:25 +01:00
David van Moolenbroek
0bb27bb0b1 Servers: remove ABI comment 2011-11-07 22:24:59 +01:00
David van Moolenbroek
b02c260ecb Miscellaneous legacy cleanup 2011-11-07 22:20:55 +01:00
Ben Gras
3fbb50397c stopgap local fstab
. /etc/fstab.local to list user-defined filesystems
	. /usr/etc/rc reads it and runs type-specific fsck if necessary
	. also type-specific mount
2011-11-07 20:37:12 +01:00
David van Moolenbroek
0f3423cd75 Kernel: fix compilation with DEBUG_RACE on 2011-11-07 19:59:02 +01:00
David van Moolenbroek
a80eb6545a blocktest: add system.conf file, too 2011-11-07 18:26:16 +01:00
David van Moolenbroek
2602861f23 Move optset.c into libsys; remove redundant copies 2011-11-07 16:16:08 +01:00
David van Moolenbroek
fc34180cca Add "blocktest" test, for testing block drivers
The test is located in test/blocktest. It requires hand-editing of
its configuration, and is not part of any automated test set.
2011-11-07 14:55:55 +01:00
David van Moolenbroek
ff542f0b27 atl2: prevent compiler memory access misordering
While no problems have been observed in practice yet, modern compilers
may reorder memory access operations, and that could lead to problems
with memory-mapped I/O typically done by drivers. This patch prevents
any potentially problematic reordering by the compiler in the ATL2
driver.

In addition, this patch removes a number of gcc/clang warnings.
2011-11-07 14:25:09 +01:00
Ben Gras
c83a90e2f1 mfs: ditch global readahead 2011-11-06 19:15:12 +01:00
David van Moolenbroek
47aad344c3 Drivers: fix issues introduced by commit b198207 2011-11-04 23:54:04 +01:00
David van Moolenbroek
5320333614 ahci: prevent compiler memory access misordering
While no problems have been observed in practice yet, modern compilers
may reorder memory access operations, and that could lead to problems
with memory-mapped I/O typically done by drivers. This patch prevents
any potentially problematic reordering by the compiler in the AHCI
driver.
2011-11-04 14:13:24 +01:00
David van Moolenbroek
e1d867b686 ahci/libdriver: multithreading support
This patch adds support for executing multiple concurrent requests on
different devices on the same AHCI controller. The libdriver library
has been extended to include a generic multithreading interface, and
the AHCI driver has been extended to make use of this interface.

The original version of this code has been written by Arne Welzel.
2011-11-04 09:37:53 +00:00
David van Moolenbroek
0812293b47 procfs: fix for PID reuse between updates
In certain cases, a process ID may be reused between two lazy updates
of procfs's PID table. If the new associated process slot has a lower
index than the old one, this will trigger an assert in vtreefs, as the
new PID name entry is added before the old one is removed. This patch
fixes the problem by always first removing old PID name entries before
adding new ones.

Bug reported by Stephen Hatton.
2011-11-01 23:50:55 +00:00
David van Moolenbroek
8b00ebde78 Kernel: remove unused MF_ASYNMSG 2011-11-01 19:21:19 +00:00
Arun Thomas
d134121b09 Install servers in GCC/clang case 2011-11-02 18:46:06 +01:00
Arun Thomas
8b4492ed1c kernel: remove unused var 2011-11-02 18:45:08 +01:00
Arun Thomas
62841e2935 pm: remove dead minix_munmap functions 2011-11-02 18:43:59 +01:00
David van Moolenbroek
1992529666 Kernel: fix for senda erroneously setting errors
Bug reported and fixed by Arne Welzel.
2011-10-31 18:43:37 +00:00
Thomas Veerman
65e54c23e3 PFS is also a system proc 2011-10-28 12:49:47 +00:00
Thomas Veerman
febb4403ff Reenable backcall detection 2011-10-27 15:29:37 +00:00
Thomas Veerman
823e42c3e3 Make AVFS deal intelligently with back calling FSes
PUFFS file systems need to make back calls for every operation we
send to them. Consequently, they cannot handle block reads and writes
themselves. Instead, the root file system has to do it (for now).

When the mount operation causes an FS to make a back call, AVFS now
concludes that every block read and write for that FS has to go
through the root file system.
2011-10-27 15:29:23 +00:00