Teach mount(8) about a 'late' keyword, which means the file system should
not be mounted unless the -l flag was specified.
Add an rc script, mountlate, which basically runs 'mount -a -l'. It runs
after DAEMON but before LOGIN.
This is useful for things like loopback mounts, because mountcritremote
runs before mountd / nfsd (since /usr might be a remote file system), so
an attempt to mount a loopback network file system in mountcritremote will
fail.
Also add a progress message to mountcritlocal, for the sake of symmetry
with similar messages in mountcritremote and mountlate.
Always supply curthread as argument to nfs_asyncio and nfs_doio
in nfs_strategy. Otherwise, for some buffers, signals would be ignored
at the intr mounts.
Matt Jacob [Wed, 12 Jul 2006 07:48:50 +0000 (07:48 +0000)]
Fix config page writes to not strip out the attributes when you
actually go write the config page. This fixes the long standing
problem about updating NVRAM on Fibre Channel cards and seems
so far to not break SPI config page writes.
Put back role setting into mpt. That is, you can set a desired role
for mpt as a hint. On the next reboot, it'll pick that up and redo
the NVRAM settings appropriately and warn you that this won't take
effect until the next reboot. This saves people the step of having
to find a BIOS utilities disk to set target and/or initiator role
for the MPT cards.
David Xu [Wed, 12 Jul 2006 05:54:17 +0000 (05:54 +0000)]
Return priority range 0..PRI_MAX_TIMESHARE-PRI_MIN_TIMESHARE for
SCHED_OTHER, the same range as rtprio() is using. In old code,
it returns nice range -20 .. 20, nice should be treated as process
weight, it is really managed by getpriority() and setpriority()
syscalls, they are different.
Jung-uk Kim [Wed, 12 Jul 2006 00:49:46 +0000 (00:49 +0000)]
Assume floppy disk is not inserted when we have exhausted retries. This
significantly reduces booting time when there is broken floppy disk drive,
controller, cable, BIOS, etc.
When the floppy controller interface is correctly implemented, disk change
signal (DSKCHG) is reflected in the Digital Input Register (DIR) at 0x3f7.
However, there are many cases that the signal is unusable. Moreover, some
BIOS does not reserve the port at all. In those cases, the register may not
function.
Comment out the mapping of the OHCI controller registers va == pa. This
address is in the userland address space. The proper thing is either to choose
a virtual address in the kernel address space beyond the KVA, or to use
pmap_mapdev().
Robert Watson [Tue, 11 Jul 2006 23:18:28 +0000 (23:18 +0000)]
Several protocol switch functions (pru_abort, pru_detach, pru_sosetlabel)
return void, so don't implement no-op versions of these functions.
Instead, consistently check if those switch pointers are NULL before
invoking them.
Robert Watson [Tue, 11 Jul 2006 21:56:58 +0000 (21:56 +0000)]
When pru_attach() fails, call sodealloc() on the socket rather than
using sorele() and the full tear-down path. Since protocol state
allocation failed, this is not required (and is arguably undesirable).
This matches the behavior of sonewconn() under the same circumstances.
Robert Watson [Tue, 11 Jul 2006 21:49:54 +0000 (21:49 +0000)]
Reduce periods of simultaneous acquisition of various socket buffer
locks and the unplock during uipc_rcvd() and uipc_send() by caching
certain values from one structure while its locks are held, and
applying them to a second structure while its locks are held. If
done carefully, this should be correct, and will reduce the amount
of work done with the global unp lock held.
John Baldwin [Tue, 11 Jul 2006 20:52:08 +0000 (20:52 +0000)]
- Add conditional VFS Giant locking to getdents_common() (linux ABIs),
ibcs2_getdents(), ibcs2_read(), ogetdirentries(), svr4_sys_getdents(),
and svr4_sys_getdents64() similar to that in getdirentries().
- Mark ibcs2_getdents(), ibcs2_read(), linux_getdents(), linux_getdents64(),
linux_readdir(), ogetdirentries(), svr4_sys_getdents(), and
svr4_sys_getdents64() MPSAFE.
The ufs_lookup.c has a critical bug around the whiteout
process. UFS must check a whiteout name when it uses the
whiteout, but the current implementation does not check
the whileout name, so sometimes UFS writes over a wrong
whtieout. UFS *MUST* check the whiteout name to use a
corrent whiteout. This bug leads unionfs. panic.
This commit fixes this trouble.
Add a new flag to pmap_enter_locked() to say if it's OK to wait. If it is, and
we're unable to allocate the memory for a PTE, we'll wait until we can. If not,
we'll just return.
Use M_NOWAIT|M_USE_RESERVE to allocate PTEs, it is less aggressive than
M_NOWAIT alone.
David Xu [Tue, 11 Jul 2006 06:11:34 +0000 (06:11 +0000)]
Extended the POSIX scheduler APIs to accept lwpid as well, we've already
done this in ptrace syscall, when a pid is large than PID_MAX, the syscall
will search a thread in current process. It permits 1:1 thread library to
get and set a thread's scheduler attributes.
David Xu [Tue, 11 Jul 2006 06:01:14 +0000 (06:01 +0000)]
For SCHED_OTHER, we always inherit current thread's interactive priority
unless current thread is realtime thread, in such case, we set a new zero
priority for it, notice we don't have per-thread nice, the priority
passed by userland is ignored here.
John Baldwin [Mon, 10 Jul 2006 21:38:17 +0000 (21:38 +0000)]
- Split out kern_accept(), kern_getpeername(), and kern_getsockname() for
use by ABI emulators.
- Alter the interface of kern_recvit() somewhat. Specifically, go ahead
and hard code UIO_USERSPACE in the uio as that's what all the callers
specify. In place, add a new uioseg to indicate what type of pointer
is in mp->msg_name. Previously it was always a userland address, but
ABI emulators may pass in kernel-side sockaddrs. Also, remove the
namelenp field and instead require the two places that used it to
explicitly copy mp->msg_namelen out to userland.
- Use the patched kern_recvit() to replace svr4_recvit() and the stock
kern_sendit() to replace svr4_sendit().
- Use kern_bind() instead of stackgap use in ti_bind().
- Use kern_getpeername() and kern_getsockname() instead of stackgap in
svr4_stream_ti_ioctl().
- Use kern_connect() instead of stackgap in svr4_do_putmsg().
- Use kern_getpeername() and kern_accept() instead of stackgap in
svr4_do_getmsg().
- Retire the stackgap from SVR4 compat as it is no longer used.
John Baldwin [Mon, 10 Jul 2006 19:28:57 +0000 (19:28 +0000)]
Explicitly use STAILQ_REMOVE_HEAD() when we know we are removing the head
element to avoid confusing Coverity. It's now also easier for humans to
parse as well.
Implement mpo_associate_nfsd_label entry point for the BIBA security policy,
we will initialize the label to biba/low for files that have been created
through an NFS RPC. This is a safe default given the default nature of our
NFS implementation, there is not a whole lot of data integrity there by
default. This also fixes kernel panics associated with file creation over NFS
while creating files on filesystems which have multilabel enabled with BIBA
enabled.
John Baldwin [Mon, 10 Jul 2006 17:59:26 +0000 (17:59 +0000)]
Retire the stackgap macros from ibcs2 as they are no longer used. Push
the includes of <sys/exec.h> and <sys/sysent.h> down into the only files
that now need them.
Extend i4b to support CAPI manager based ISDN controllers (CAPI manager is part of
c4b, CAPI for BSD). This is a preparation to add CAPI for BSD to the source tree.
Matt Jacob [Sun, 9 Jul 2006 17:50:20 +0000 (17:50 +0000)]
Convert isp(4) and ispfw(4) to use firmware(9) to manage firmware
loading for the QLogic cards.
Because isp(4) exists before the root is mounted, it's not really
possible for us to use the kernel's linker to load modules directly
from disk- that's really too bad.
However, the this is still a net win in in that the firmware has
been split up on a per chip (and in some cases, functionality)
basis, so the amount of stuff loaded *can* be substantially less
than the 1.5MB of firmware images that ispfw now manages. That is,
each specific f/w set is now also built as a module. For example,
QLogic 2322 f/w is built as isp_2322.ko and Initiator/Target 1080
firmware is built as isp_1080_it.ko.
For compatibility purposes (i.e., to perturb folks the least), we
also still build all of the firmware as one ispfw.ko module.
This allows us to let 'ispfw_LOAD' keep on working in existing
loader.conf files. If you now want to strip this down to just
the firmware for your h/w, you can then change loader.conf to
load the f/w you specifically want.
We also still allow for ispfw to be statically built (e.g., for
PAE and sparc64).
Future changes will look at f/w unloading and also role switching
that then uses the kernel linker to load different ips f/w sets.
MFC after: 2 months
Change fs->fs_fsmnt to mp->mnt_stat.f_mntonname in warnings about missing
MAC and ACLs support in the kernel. If it is a first mount, fs->fs_fsmnt
is empty.
When kern.geom.raid3.use_malloc tunnable is set to 1, malloc(9) instead of
uma(9) will be used for memory allocation.
In case of problems or tracking bugs, there are more useful tools for malloc(9)
debugging than for uma(9) debugging, like memguard(9) and redzone(9).
Sam Leffler [Sun, 9 Jul 2006 06:10:23 +0000 (06:10 +0000)]
o replace special handling of clone operations by a clone callback
mechanism
o change vlan cloning to use callback and pass all vlan parameters
on create using the new SIOCREATE2 ioctl
o update vlan set logic to match existing practice
Sam Leffler [Sun, 9 Jul 2006 06:04:01 +0000 (06:04 +0000)]
Revise network interface cloning to take an optional opaque
parameter that can specify configuration parameters:
o rev cloner api's to add optional parameter block
o add SIOCCREATE2 that accepts parameter data
o rev vlan support to use new api (maintain old code)
Colin Percival [Sun, 9 Jul 2006 05:53:24 +0000 (05:53 +0000)]
Remove build timestamps. There's no need to know exactly when these programs
were built, and this removal makes these binaries build the same every time.
John Baldwin [Sat, 8 Jul 2006 20:12:14 +0000 (20:12 +0000)]
- Split ioctl() up into ioctl() and kern_ioctl(). The kern_ioctl() assumes
that the 'data' pointer is already setup to point to a valid KVM buffer
or contains the copied-in data from userland as appropriate (ioctl(2)
still does this). kern_ioctl() takes care of looking up a file pointer,
implementing FIONCLEX and FIOCLEX, and calling fi_ioctl().
- Use kern_ioctl() to implement xenix_rdchk() instead of using the stackgap
and mark xenix_rdchk() MPSAFE.
John Baldwin [Sat, 8 Jul 2006 20:05:04 +0000 (20:05 +0000)]
Use kern_connect() in spx_open() to avoid the need for the stackgap. I
also used kern_close() for simplicity though close(2) wasn't requiring
the use of the stackgap.
John Baldwin [Sat, 8 Jul 2006 19:54:12 +0000 (19:54 +0000)]
- Split the IBCS2 ipc foosys() system calls up into subfunctions matching
the organization in svr4_ipc.c.
- Use kern_msgctl(), kern_semctl(), and kern_shmctl() instead of the
stackgap.
John Baldwin [Sat, 8 Jul 2006 19:51:38 +0000 (19:51 +0000)]
Rework kern_semctl a bit to always assume the UIO_SYSSPACE case. This
mostly consists of pushing a few copyin's and copyout's up into
__semctl() as all the other callers were already doing the UIO_SYSSPACE
case. This also changes kern_semctl() to set the return value in a passed
in pointer to a register_t rather than td->td_retval[0] directly so that
callers can only set td->td_retval[0] if all the various copyout's succeed.
As a result of these changes, kern_semctl() no longer does copyin/copyout
(except for GETALL/SETALL) so simplify the locking to acquire the semakptr
mutex before the MAC check and hold it all the way until the end of the
big switch statement. The GETALL/SETALL cases have to temporarily drop it
while they do copyin/malloc and copyout. Also, simplify the SETALL case to
remove handling for a non-existent race condition.
Warner Losh [Sat, 8 Jul 2006 17:06:15 +0000 (17:06 +0000)]
Create bus_enumerate_hinted_children. This routine will allow drivers
to use the hinted child system. Bus drivers that use this need to
implmenet the bus_hinted_child method, where they actually add the
child to their bus, as they see fit. The bus is repsonsible for
getting the attribtues for the child, adding it in the right order,
etc. ISA hinting will be updated to use this method.
Warner Losh [Sat, 8 Jul 2006 15:51:55 +0000 (15:51 +0000)]
Remove old GENERIC kludge. We no longer need to skip devices named
atkbd. Version 1.162 of GENERIC fixed this problem in April of 1999.
Subsequent to that, the hints data was removed from GENERIC and move
to hints files. All the hints file ever created have atkbd at the
right location. This should have been removed just after RELENG_4 was
branched (and likely around 4.5 in RELENG_4).
Always supply curthread as argument to nfs_asyncio and nfs_doio
in nfs_strategy. Otherwise, for some buffers, signals would be ignored
at the intr mounts.