Commit graph

1425 commits

Author SHA1 Message Date
Lionel Sambuc
7d98eee859 usr.bin/make: -j fix
The job option enable the usage of pipes to communicate with sub-makes.

On MINIX, it seems that there is possibility of receiving an EAGAIN on
such reads, even when it had previously been tested for POLLIN using
poll().

This patch is a workaround, by wrapping the read operation within a
do {} while(errno == EAGAIN && ...) loop.

Change-Id: Ia184c4d600efe7218d197820df87761604120862
2014-07-28 17:05:15 +02:00
Lionel Sambuc
ac9fa699bd Fix perl build hanging on pipe
Change-Id: Ic1ff95f45c6082b9310b27783190ce8f92c2afad
2014-07-28 17:05:14 +02:00
Ben Gras
013226c417 vfs: call pipe_check before reviving select()
. select callback on close() on a pipe
	  would falsely revive a read-select on
	  the pipe
	. the extra pipe_check catches it
	. fixes the new import of the netbsd
	  make in -j N mode (make jobs that communicate
	  with pipes)
2014-07-28 17:05:14 +02:00
Ben Gras
58b8ff5ffa VFS: add F_FLUSH_FS_CACHE fcntl
This fcntl requests all cached blocks associated with the minor device
number associated with the regular file are invalidated. If the file
is a block special, invalidate the blocks associated with that minor
device instead.

This is to be used for a test that tests unmapped file-mapped memory
ranges whose blocks are not in the cache and therefore must be fetched
from a FS.

Change-Id: Ide914b2e88413aa90bd731ae587ca06fa5f13ebc
2014-07-28 17:05:14 +02:00
Ben Gras
565f13088f make vfs & filesystems use failable copying
Change the kernel to add features to vircopy and safecopies so that
transparent copy fixing won't happen to avoid deadlocks, and such copies
fail with EFAULT.

Transparently making copying work from filesystems (as normally done by
the kernel & VM when copying fails because of missing/readonly memory)
is problematic as it can happen that, for file-mapped ranges, that that
same filesystem that is blocked on the copy request is needed to satisfy
the memory range, leading to deadlock. Dito for VFS itself, if done with
a blocking call.

This change makes the copying done from a filesystem fail in such cases
with EFAULT by VFS adding the CPF_TRY flag to the grants. If a FS call
fails with EFAULT, VFS will then request the range to be made available
to VM after the FS is unblocked, allowing it to be used to satisfy the
range if need be in another VFS thread.

Similarly, for datacopies that VFS itself does, it uses the failable
vircopy variant and callers use a wrapper that talk to VM if necessary
to get the copy to work.

	. kernel: add CPF_TRY flag to safecopies
	. kernel: only request writable ranges to VM for the
	  target buffer when copying fails
	. do copying in VFS TRY-first
	. some fixes in VM to build SANITYCHECK mode
	. add regression test for the cases where
	  - a FS system call needs memory mapped in a process that the
	    FS itself must map.
	  - such a range covers more than one file-mapped region.
	. add 'try' mode to vircopy, physcopy
	. add flags field to copy kernel call messages
	. if CP_FLAG_TRY is set, do not transparently try
	  to fix memory ranges
	. for use by VFS when accessing user buffers to avoid
	  deadlock
	. remove some obsolete backwards compatability assignments
        . VFS: let thread scheduling work for VM requests too
          Allows VFS to make calls to VM while suspending and resuming
          the currently running thread. Does currently not work for the
          main thread.
        . VM: add fix memory range call for use by VFS

Change-Id: I295794269cea51a3163519a9cfe5901301d90b32
2014-07-28 17:05:14 +02:00
Ben Gras
8c277f99f4 PFS: when copy fails, don't change state
. makes it safe to transparently repeate i/o operation on
	  EFAULT
	. needed for future re-issued i/o operation change in VFS

Change-Id: Iea6b29f3bff19002dcd69b30569671768ac17e21
2014-07-28 17:05:13 +02:00
David van Moolenbroek
b539f91aba RS: also wait for VM's RS_INIT message
Bug reported by JD Begin.

Change-Id: Idd8c4299f20253fe2d982af319002bfb67af27c6
2014-07-28 17:05:13 +02:00
David van Moolenbroek
f199fc0bfe VM: fix corruption from recursive PDE allocation
Change-Id: I6176b849fefca4bed3e92648b0d72ff47658915c
2014-07-28 17:05:13 +02:00
Ben Gras
273c51a7d1 64-bit CDEV_POS
Change-Id: Iecdadf3dd92cc14525961976f51a9eb16a996d62
2014-07-28 17:05:11 +02:00
Ben Gras
3f3e78ef59 64-bit REQ_TRC_START, REQ_TRC_END
Change-Id: I567804209695b5ec0d83a453d93e36cbf8900f1a
2014-07-28 17:05:11 +02:00
Ben Gras
e6daf20a69 64-bit VFS_LSEEK_OFF
Change-Id: Ic0b6d65cbde1033462b909436efa92464094f1ec
2014-07-28 17:05:11 +02:00
Ben Gras
ed9076ccb4 64-bit VFS_VMCALL_OFFSET
Change-Id: I29725365a199f850420cd0e4e3902cf70dffe9ad
2014-07-28 17:05:10 +02:00
Ben Gras
188a5fde1f 64-bit VFS_TRUNCATE_OFF
Change-Id: I4bd5cc57ddda2525b0bec6f044f35196a2c21f2e
2014-07-28 17:05:10 +02:00
Ben Gras
4694fcc1d5 64-bit RES_FILE_SIZE
Change-Id: Ia17b4b4722f87fb35142948b62d4a8392ded5f73
2014-07-28 17:05:10 +02:00
Ben Gras
fdd85c4d08 64-bit RES_SEEK_POS
Change-Id: Id13f06417f2f600db167bec7b33bc825742cae79
2014-07-28 17:05:10 +02:00
Ben Gras
978082bb0d 64-bit REQ_SEEK_POS
Change-Id: I2e51a188b171af0f0a02349e4eccbe78e7cc2e0c
2014-07-28 17:05:10 +02:00
Lionel Sambuc
d3d33afe9f Experimental pthread compatibility library
This patch adds pthread compatibility by using libmthread.

To use this with a program using pthreads, you have to replace
  #include <pthread>
with
  #define _MTHREADIFY_PTHREADS
  #include <minix/mthreads>

This also changes the initialization function to be a constructor, which
is implicitly called before the call to main. This allows for
conformance with pthreads, while not paying a high price by checking on
each mthread_* call whether the library has been initialized or not.

As mthread_init is now a constructor, it also has been set as static, and
relevent calls removed from programs using it.

Change-Id: I2aa375db557958d2bee9a70d285aabb990c88f00
2014-07-28 17:05:08 +02:00
Lionel Sambuc
84d9c625bf Synchronize on NetBSD-CVS (2013/12/1 12:00:00 UTC)
- Fix for possible unset uid/gid in toproto
 - Fix for default mtree style
 - Update libelf
 - Importing libexecinfo
 - Resynchronize GCC, mpc, gmp, mpfr
 - build.sh: Replace params with show-params.
     This has been done as the make target has been renamed in the same
     way, while a new target named params has been added. This new
     target generates a file containing all the parameters, instead of
     printing it on the console.
 - Update test48 with new etc/services (Fix by Ben Gras <ben@minix3.org)
     get getservbyport() out of the inner loop

Change-Id: Ie6ad5226fa2621ff9f0dee8782ea48f9443d2091
2014-07-28 17:05:06 +02:00
Ben Gras
aae07bc777 <sys/ioccom.h>, <sys/ioctl.h>
. use <sys/ioccom.h> to make ioctls
	. use netbsd <sys/ioctl.h>, include minix ioctls
	. convert to varargs ioctl() like netbsd

Change-Id: Id5165780df48172b28bf0003603516d30d7c1abb
2014-03-03 20:47:06 +01:00
Ben Gras
a1e1ae5ade <sys/fcntl.h>, remove O_REOPEN
Change-Id: Id6493f30bb0fec7460d01f357ada9b2c2a4f5eb7
2014-03-03 20:47:06 +01:00
Ben Gras
be9fe09e97 x86 multiboot.h
Change-Id: I245564a98fb9e2572b88f8feb7411ad6800a543c
2014-03-03 20:47:05 +01:00
Ben Gras
17587738d3 align <sys/ucontext.h> <sys/uio.h> <sys/un.h>
Change-Id: I70adf01fddf931a3a6931083adaa4bbe647ea6a3
2014-03-03 20:47:03 +01:00
Ben Gras
dda632a24f drop the minix_ prefixes for mmap and munmap
also cleanup of various minix-specific changes, cleanup of
mmap-related testing.

Change-Id: I289a4fc50cf8a13df4a6082038d860853a4bd024
2014-03-03 20:47:03 +01:00
Lionel Sambuc
3053b27247 include/limits.h: merged
Replacing SYMLOOP_MAX, SYMLINK_MAX, by _POSIX_SYMLOOP_MAX,
_POSIX_SYMLINK_MAX respectively.

Change-Id: I020c20b0dbf6d44e777879d786a20811f23d1561
2014-03-03 20:47:03 +01:00
Lionel Sambuc
b33edbfe7a Removing VFS_UTIMENS_* field macros.
Change-Id: I3be1acfbb77b6e229d784d8dbc3ceec1b26abf90
2014-03-03 20:47:00 +01:00
Lionel Sambuc
301f5f87f0 Renamed m_vm_vfs to m_vm_vfs_mmap.
Stay coherent with the naming scheme of the messages.

Change-Id: Icc0e13a88ec29263502166c0e6eec81cdb974663
2014-03-03 20:47:00 +01:00
Lionel Sambuc
175d3e7eae Changing the message union to anonymous.
This allows us to write things like this:
  message m;
  m.m_notify.interrupts = new_value;

or
  message *mp;
  mp->m_notify.interrupts = new_value;

The shorthands macro have been adapted for the new scheme, and will be
kept as long as we have generic messages being used.

Change-Id: Icfd02b5f126892b1d5d2cebe8c8fb02b180000f7
2014-03-03 20:46:47 +01:00
Lionel Sambuc
4a0199d66d PM_TIME_SEC into 64bit
Change-Id: I609ee8cac7aae8af2c0c0381710433a5df0caebd
2014-03-03 20:45:28 +01:00
Lionel Sambuc
55d9e9954b Extending time_t to 64bits.
Change-Id: Ia96b8bfba19cb8179a0237a7d2122d415c24d73f
2014-03-03 20:45:28 +01:00
Ben Gras
88be7bd333 Use netbsd <sys/mman.h>
Change-Id: I80e9cffc80140383a6faf692248573c64d282b4a
2014-03-03 20:37:27 +01:00
Ben Gras
256829c7e8 <sys/wait.h>
Change-Id: I24605df209d012e99333065428a3ed2752799321
2014-03-02 12:28:32 +01:00
Gerard
78da142dab Replaced common u64.h functions with operators.
Change-Id: I71b7b4879209eeff89ce5748d67102afebf871dc
2014-03-02 12:28:32 +01:00
Lionel Sambuc
37598dccf1 Aligning dev_t to 64bits.
Change-Id: I630f72f8530dd4aaf05c35ca23683ae12c9f8328
2014-03-02 12:28:32 +01:00
Ben Gras
8ccb12bb5a use netbsd <sys/signal.h> and sigset_t
. create signals-related struct message type to store sigset_t
	  directly
	. create notify-specific message types, so the generic NOTIFY_ARG
	  doesn't exist anymore
	. various related test expansions, improvements, fixes
	. add a few error-checks to sigismember() calls
	. rename kernel call specific signals fields to SYS_*

Change-Id: I53c18999b5eaf0cfa0cb25f5330bee9e7ad2b478
2014-03-02 12:28:31 +01:00
Ben Gras
25719b5d92 bigger message
Change-Id: Ie770140c55799bdc3bb8f0ad6994d59938155a1a
2014-03-02 12:28:31 +01:00
Ben Gras
7120f34ec1 drop <minix/termios.h>, use clean <sys/termios.h>
. also implement some netbsd-style tty ioctls
	. also implement SIGINFO
	. also import netbsd stty
	. rename keymap minix CMIN (for ctrl+minus on numeric keypad)
	  to CNMIN; to keep unchanged control character default CMIN in
	  new <sys/ttydefaults.h>
	. convert CS[5678] logic in rs232 driver to explicit setting of LC
	  bits

Change-Id: I9b7d2963fe9aec00fb6e7535ef565b3191fc1c1d
2014-03-02 12:28:20 +01:00
Ben Gras
a06e2ab395 big <utmp.h>-inspired netbsd switch
import/switch of:
init, getty, reboot, halt, shutdown, wall, last

changes:
	. change reboot() call to netbsd prototype and args
	. allows pristine <utmp.h>
	. use clean <sys/reboot.h> instead of <minix/reboot.h>
	. implement TIOCSCTTY for use by getty so getty can get
	  controlling terminal from init's child(ren)
	. allow NULL envp for exec

Change-Id: I5ca02cb4230857140c08794bbfeba7df982c58a3
2014-03-01 09:05:02 +01:00
Lionel Sambuc
c3fc9df84a Adding ipc_ prefix to ipc primitives
* Also change _orig to _intr for clarity
 * Cleaned up {IPC,KER}VEC
 * Renamed _minix_kernel_info_struct to get_minix_kerninfo
 * Merged _senda.S into _ipc.S
 * Moved into separate files get_minix_kerninfo and _do_kernel_call
 * Adapted do_kernel_call to follow same _ convention as ipc functions
 * Drop patches in libc/net/send.c and libc/include/namespace.h

Change-Id: If4ea21ecb65435170d7d87de6c826328e84c18d0
2014-03-01 09:05:01 +01:00
David van Moolenbroek
24ec0d73b5 Clean up interface to PM and VFS
- introduce new call numbers, names, and field aliases;
- initialize request messages to zero for all ABI calls;
- format callnr.h in the same way as com.h;
- redo call tables in both servers;
- remove param.h namespace pollution in the servers;
- make brk(2) go to VM directly, rather than through PM;
- remove obsolete BRK, UTIME, and WAIT calls;
- clean up path copying routine in VFS;
- move remaining system calls from libminlib to libc;
- correct some errno-related mistakes in libc routines.

Change-Id: I2d8ec5d061cd7e0b30c51ffd77aa72ebf84e2565
2014-03-01 09:05:01 +01:00
David van Moolenbroek
2c8310fce6 VFS: store m_out as part of worker thread state
There is no need to pass pointers around when there is a structure
available that already stores other similar state, such as m_in.

Change-Id: I3164c5c55c71f443688103d1f0756c086eb05974
2014-03-01 09:05:00 +01:00
David van Moolenbroek
d8106f7f1d Rename VFS_BASE to FS_BASE
The recipients of these requests are file systems, not VFS.

Change-Id: I49211cfe4289df25b0c07171c7f8c321fc3731d6
2014-03-01 09:05:00 +01:00
David van Moolenbroek
87c599da2d PM/VFS protocol: prefix with VFS_PM_
These calls are sent to VFS, and thus should be prefixed with VFS_.
Clean up the protocol and PM's main function a bit.

Since the protocol is substantially big and different from normal VFS
requests, this protocol retains its own numbering range for now.

Change-Id: Ia62104b5c5c929ed787144816d2e4cc70bed3b0b
2014-03-01 09:05:00 +01:00
David van Moolenbroek
44d3230e40 For common calls, give servers unique call numbers
The getsysinfo(2), getrusage(2), and svrctl(2) calls used the same
call number to different services. Since we want to give each service
its own call number ranges, this is no longer tenable. This patch
introduces per-service call numbers for these calls.

Note that the remainder of the COMMON_ range is left intact, as these
the remaining requests in it are processed by SEF and thus server-
agnostic. The range should really be prefixed with SEF_ now.

Change-Id: I80d728bbeb98227359c525494c433965b40fefc3
2014-03-01 09:05:00 +01:00
David van Moolenbroek
80bd109cd3 libsys: various updates
- move system calls for use by services from libminlib into libsys;
- move srv_fork(2) and srv_kill(2) from RS and into libsys;
- replace getprocnr(2) with sef_self(3);
- rename previous getnprocnr(2) to getprocnr(2);
- clean up getepinfo(2);
- change all libsys calls that used _syscall to use _taskcall, so as
  to avoid going through errno to pass errors; this is already how
  most calls work anyway, and many of the calls previously using
  _syscall were already assumed to return the actual error;
- initialize request messages to zero, for future compatibility
  (note that this does not include PCI calls, which are in need of a
  much bigger overhaul, nor kernel calls);
- clean up more of dead DS code as a side effect.

Change-Id: I8788f54c68598fcf58e23486e270c2d749780ebb
2014-03-01 09:05:00 +01:00
David van Moolenbroek
efd3487bc5 PM: send replies immediately
The original delayed reply functionality was there to support swapping
in processes as they are unblocked, but swap support is long gone.
These days, this code only incurs overhead and hides bugs.

Change-Id: I4aebcd80719daa1bec45ac91975ddc9a460d74d4
2014-03-01 09:05:00 +01:00
David van Moolenbroek
f310aefcbd PM: resolve fork/kill race condition
When a process forks, VFS is informed on behalf of the child. This is
correct, because otherwise signals to the new child could get lost.
However, that means that the parent is not blocked from being killed
by a signal while the child is blocked on this VFS call. As a result,
by the time that the VFS reply comes in, the parent may already be
dead, and the child may thus have been assigned a new parent: INIT.

Previously, PM would blindly reply to the parent when the VFS reply
for the fork came in. Thus, it could end up sending a reply to INIT,
even though INIT did not issue the fork(2) call. This could end up
satisfying a different call from INIT (typically waitpid(2)) and then
cause an error when that other call was complete.

It would be possible to set VFS_CALL on both forking parent and child.
This patch instead adds a flag (NEW_PARENT) to note that a process's
parent has changed during a VFS call.

Change-Id: Iad930b2e441db54fe6f7d2fd011f0f6a26e2923d
2014-03-01 09:04:59 +01:00
David van Moolenbroek
595d73a896 PM: rework signal handling
- introduce PROC_STOPPED flag, which tracks whether the process is
  stopped on PROC_STOP in the kernel, rather than implicitly deriving
  this from PM_SIG_PENDING;
- make the process resumption test based on current state rather than
  state transitions;
- add and clarify several flag checks in the signal handling code;
- add test79 to test signal handling robustness.

Change-Id: Ic8c7527095035b300b56f2ab1b9dd190bd4bf001
2014-03-01 09:04:59 +01:00
David van Moolenbroek
4d4e70daaa PM: remove obsolete ALARM, PAUSE, TIMES calls
NetBSD libc implements these as wrappers around setitimer(2),
sigsuspend(2), and getrusage(2), respectively.

Change-Id: I0c5e725b3e1316bddd3a3ff7ef65d57d30afd10d
2014-03-01 09:04:59 +01:00
David van Moolenbroek
f30a16b159 Remove support for MKTRACE, MKMCONTEXT, MKSTATECTL
Change-Id: Ib5fa53913ecb7b46d30d391dbdd3e8ef21eb2254
2014-03-01 09:04:59 +01:00
Ben Gras
6c8f7fc3ba netbsd dirent.h, import sys/sys *.h, mfs cleanup
. add all sys/sys headers not already present to help compiling
	. take netbsd dirent.h and struct dirent; main result is
	  introducing d_type and d_namlen that have to be set by getdents()
	  in all FS code implementing it
	. d_off is gone
	. alignment of the struct has become 8 bytes instead of 4
	. remove _MAX_BLOCK_SIZE, _MIN_BLOCK_SIZE, _STATIC_BLOCK_SIZE
	. libminlib: cleanup unused yet duplicate code
	. mfs: throw out the long-broken v1, v2 support
	. new test for dirent contents filled by getdents()

Change-Id: I1459755c7ba5e5d1c9396d3a587ce6e63ddc283e
2014-03-01 09:04:59 +01:00