It is possible for multiple process to sleep concurrently waiting
for a buffer. When the buffer shortage is a shortage of space but
not a shortage of buffer headers, the processes took turns creating
empty buffers and waking each other to advertise the brelse() of
the empties; progress was never made because tsleep() always found
another high-priority process to run and everything was done at
splbio(), so vfs_update never had a chance to flush delayed writes,
not to mention that i/o never had a chance to complete.
The problem seems to be rare in practice, but it can easily be
reproduced by misusing block devices, at least for sufficently slow
devices on machines with a sufficiently small buffer cache. E.g.,
`tar cvf /dev/fd0 /kernel' on an 8MB system with no disk in fd0
causes the problem quickly; the same command with a disk in fd0
causes the problem not quite as quickly; and people have reported
problems newfs'ing file systems on block devices.
Block devices only cause this problem indirectly. They are pessimized
for time and space, and the space pessimization causes the shortage
(it manifests as internal fragmentation in buffer_map).
This should be fixed in 2.2.
cost since it is only done in cpu_switch(), not for every exception.
The extra state is kept in the pcb, and handled much like the npx state,
with similar deficiencies (the state is not preserved across signal
handlers, and error handling loses state).
shared function.
- use p->p_sleepend to try and get more accurate "time remaining" results
when the time has been adjusted.
- verify writeability of return address so that we can fail before sleeping
if the address for the result is bogus.
Changes to pmap.c for lapic_t lapic && ioapic_t ioapic pointers,
currently equal to apic_base && io_apic_base, will stand alone with the
private page mapping.
be (eventually) architecture independent. It provides an emulation
of the ISA interrupt registration function register_intr(), but that
function does no longer manipulated the interrupt controller and
interrupt descriptor table, but calls the architecture dependent
function setup_icu() for that purpose.
After theISA/EISA bus code has been modified to directly call the new
interrupt registartion functions (intr_create() and intr_connect()),
the emulation of register_intr() should be dropped.
The C level interrupt handler function should take a (void*) argument,
and the function pointer type (inthand2_t) should defined in some other
place than isa_device.h.
This commit is a pre-requisite for the removal of the PCI specific shared
interrupt code.
Reviewed by: dfr,bde
This is now the default, it delays most of the MP startup to the function
machdep.c:cpu_startup(). It should be possible to move the 2 functions
found there (mp_start() & mp_announce()) even further down the path once
we know exactly where that should be...
Help from: Peter Wemm <peter@spinner.dialix.com.au>
- The 1st (preparse_mp_table()) counts the number of cpus, busses, etc. and
records the LOCAL and IO APIC addresses.
- The 2nd pass (parse_mp_table()) does the actual parsing of info and recording
into the incore MP table.
This will allow us to defer the 2nd pass untill malloc() & private pages
are available (but thats for another day!).
When a panic occurs early in the SMP boot process 'cpunumber()' hangs,
causing the panic string to be lost. Now the system appears to hang
in 'breakpoint()', but at least the user sees the panic string before the
hang.
switch. I needed 'LINT' to compile for other reasons so I kinda got the
blood on my hands. Note: I don't know how to test this, I don't know if
it works correctly.
panic( "xxxxx\n" );
to:
printf( "xxxxx\n" );
panic( "\n" );
For some as yet undetermined reason the argument to panic() is often NOT
printed, and the system sometimes hangs before reaching the panic printout.
So we hopefully at least print some useful info before the hang, as oppossed to
leaving the user clueless as to what has happened.
and b_validend. The changes to vfs_bio.c are a bit ugly but hopefully
can be tidied up later by a slight redesign.
PR: kern/2573, kern/2754, kern/3046 (possibly)
Reviewed by: dyson
to fill in the nfs_diskless structure, at the cost of some kernel
bloat. The advantage is that this code works on a wider range of
network adapters than netboot. Several new kernel options are
documented in LINT.
Obtained from: parts of the code comes from NetBSD.
Serious:
- An important timevalfix() in settime[ofday]() was lost.
Not so serious:
- There was a race initializing `delta' in the check for setting the
time backwards.
- The `#ifdef notyet' check for setting the time more than a day forwards
was back to front.
[[I deleted the code, it's useless because of iteration - Peter]]
- The timespec was not checked for validity in clock_settime().
- The timespec was not fully checked for validity in nanotime(). The
check in itimerfix() is too late, since the conversion from a timespec
to a timeval may overflow.
- A garbage timeval was checked in settimeofday() for the (uap->tv == NULL
&& uap->tzp != NULL) case. I added the broken check this some time ago.
Cosmetic:
- The "inadvertantly (sic) sleeping forever" test always failed. hzto()
always returns >= 1.
- The style wasn't very KNFish. (I only changed new code.)
Submitted by: bde
in NetBSD. The core of settimeofday() is moved to a seperate static
function settime() which both clock_settime() and settimeofday() call.
Note that I picked up the securelevel > 1 check from NetBSD that prevents
the clock being set backwards in high securelevel mode (this was a hole
that allowed resetting of inode access timestamps to arbitary values)
Obtained from: mostly from NetBSD, but the settime() function is from
our gettimeofday(), some tweaks by me.
the patches in freefall:/home/dfr/ld.diffs to your ld sources and set
BINFORMAT to aoutkld when linking the kernel.
Library changes and userland utilities will appear in a later commit.
".." vnode. This is cheaper storagewise than keeping it in the
namecache, and it makes more sense since it's a 1:1 mapping.
2. Also handle the case of "." more intelligently rather than stuff
the namecache with pointless entries.
3. Add two lists to the vnode and hang namecache entries which go from
or to this vnode. When cleaning a vnode, delete all namecache
entries it invalidates.
4. Never reuse namecache enties, malloc new ones when we need it, free
old ones when they die. No longer a hard limit on how many we can
have.
5. Remove the upper limit on namelength of namecache entries.
6. Make a global list for negative namecache entries, limit their number
to a sysctl'able (debug.ncnegfactor) fraction of the total namecache.
Currently the default fraction is 1/16th. (Suggestions for better
default wanted!)
7. Assign v_id correctly in the face of 32bit rollover.
8. Remove the LRU list for namecache entries, not needed. Remove the
#ifdef NCH_STATISTICS stuff, it's not needed either.
9. Use the vnode freelist as a true LRU list, also for namecache accesses.
10. Reuse vnodes more aggresively but also more selectively, if we can't
reuse, malloc a new one. There is no longer a hard limit on their
number, they grow to the point where we don't reuse potentially
usable vnodes. A vnode will not get recycled if still has pages in
core or if it is the source of namecache entries (Yes, this does
indeed work :-) "." and ".." are not namecache entries any longer...)
11. Do not overload the v_id field in namecache entries with whiteout
information, use a char sized flags field instead, so we can get
rid of the vpid and v_id fields from the namecache struct. Since
we're linked to the vnodes and purged when they're cleaned, we don't
have to check the v_id any more.
12. NFS knew about the limitation on name length in the namecache, it
shouldn't and doesn't now.
Bugs:
The namecache statistics no longer includes the hits for ".."
and "." hits.
Performance impact:
Generally in the +/- 0.5% for "normal" workstations, but
I hope this will allow the system to be selftuning over a
bigger range of "special" applications. The case where
RAM is available but unused for cache because we don't have
any vnodes should be gone.
Future work:
Straighten out the namecache statistics.
"desiredvnodes" is still used to (bogusly ?) size hash
tables in the filesystems.
I have still to find a way to safely free unused vnodes
back so their number can shrink when not needed.
There is a few uses of the v_id field left in the filesystems,
scheduled for demolition at a later time.
Maybe a one slot cache for unused namecache entries should
be implemented to decrease the malloc/free frequency.
but now that we've widened the scope of the smp work to -current, it might
be an idea to warn new people that might not have read all the docs yet
that the SMP support needs to be activated via a sysctl.
This code re-numbers PCI busses in the MP table to match PCI semantics
when the MP BIOS fails to do it properly.
Reviewed by: Peter Wemm <peter@spinner.DIALix.COM>
replace invldebug with invltlb_ok for throttling smp_invltlb() during boot.
Reviewed by: informal discussion with Peter Wemm <peter@spinner.DIALix.COM>
Peter Wemm <peter@spinner.DIALix.COM>, Steve Passe <smp@csn.net>
removed all the IPI_INTS code.
made the XFAST_IPI32 code default, renaming Xfastipi32 to Xinvltlb.
This commit includes the following changes:
1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility
glue for them is deleted, and the kernel will panic on boot if any are compiled
in.
2) Certain protocol entry points are modified to take a process structure,
so they they can easily tell whether or not it is possible to sleep, and
also to access credentials.
3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt()
call. Protocols should use the process pointer they are now passed.
4) The PF_LOCAL and PF_ROUTE families have been updated to use the new
style, as has the `raw' skeleton family.
5) PF_LOCAL sockets now obey the process's umask when creating a socket
in the filesystem.
As a result, LINT is now broken. I'm hoping that some enterprising hacker
with a bit more time will either make the broken bits work (should be
easy for netipx) or dike them out.
There are various options documented in i386/conf/LINT, there is more to
come over the next few days.
The kernel should run pretty much "as before" without the options to
activate SMP mode.
There are a handful of known "loose ends" that need to be fixed, but
have been put off since the SMP kernel is in a moderately good condition
at the moment.
This commit is the result of the tinkering and testing over the last 14
months by many people. A special thanks to Steve Passe for implementing
the APIC code!
Fix another bug: if argv[0] is NULL, garbadge args might be added for
shell script
Submitted by: Tor Egge <Tor.Egge@idi.ntnu.no> (with yet one fault detect from me)
difference of approx 3mins in make world on my P6!!! This means
that vfork now has full address space sharing, so beware with
sloppy vfork programming. Also, you really do need to apply
the previously committed popen fix in libc.
Zero the b_dirty{off,end} after cluster-comitting a group of buffers.
With these fixes, I was able to complete a 'make world' with remote src
and obj directories.
were always in a tss; that tss just changed from the one in the
pcb to common_tss (who knows where it was when there was no curpcb?).
Not using the pcb also fixed the problem that there is no pcb in
idle(), so we now always get useful register values.
cache queue more often. The pageout daemon had to be waken up
more often than necessary since pages were not put on the
cache queue, when they should have been.
Submitted by: David Greenman <dg@freebsd.org>
fork. (On my machine, fork is about 240usecs, vfork is 78usecs.)
Implement rfork(!RFPROC !RFMEM), which allows a thread to divorce its memory
from the other threads of a group.
Implement rfork(!RFPROC RFCFDG), which closes all file descriptors, eliminating
possible existing shares with other threads/processes.
Implement rfork(!RFPROC RFFDG), which divorces the file descriptors for a
thread from the rest of the group.
Fix the case where a thread does an exec. It is almost nonsense for a thread
to modify the other threads address space by an exec, so we
now automatically divorce the address space before modifying it.
longer has anything to do with vnodes and never had anything to do
with buffers, but it needs the definitions of B_READ and B_WRITE
for use with the bogus useracc() interface and was getting them
bogusly due to excessive cleanups in rev.1.49.
space. (!)
Have each process use the kernel stack and pcb in the kvm space. Since
the stacks are at a different address, we cannot copy the stack at fork()
and allow the child to return up through the function call tree to return
to user mode - create a new execution context and have the new process
begin executing from cpu_switch() and go to user mode directly.
In theory this should speed up fork a bit.
Context switch the tss_esp0 pointer in the common tss. This is a lot
simpler since than swithching the gdt[GPROC0_SEL].sd.sd_base pointer
to each process's tss since the esp0 pointer is a 32 bit pointer, and the
sd_base setting is split into three different bit sections at non-aligned
boundaries and requires a lot of twiddling to reset.
The 8K of memory at the top of the process space is now empty, and unmapped
(and unmappable, it's higher than VM_MAXUSER_ADDRESS).
Simplity the pmap code to manage process contexts, we no longer have to
double map the UPAGES, this simplifies and should measuably speed up fork().
The following parts came from John Dyson:
Set PG_G on the UPAGES that are now in kernel context, and invalidate
them when swapping them out.
Move the upages object (upobj) from the vmspace to the proc structure.
Now that the UPAGES (pcb and kernel stack) are out of user space, make
rfork(..RFMEM..) do what was intended by sharing the vmspace
entirely via reference counting rather than simply inheriting the mappings.
convenient and makes life difficult for my next commit. We still need
an i386tss to point to for the tss slot in the gdt, so we use a common
tss shared between all processes.
Note that this is going to break debugging until this series of commits
is finished. core dumps will change again too. :-( we really need
a more modern core dump format that doesn't depend on the pcb/upages.
This change makes VM86 mode harder, but the following commits will remove
a lot of constraints for the VM86 system, including the possibility of
extending the pcb for an IO port map etc.
Obtained from: bde
Use the name argument almost the same in all LKM types. Maintain
the current behavior for the external (e.g., modstat) name for DEV,
EXEC, and MISC types being #name ## "_mod" and SYCALL and VFS only
#name. This is a candidate for change and I vote just the name without
the "_mod".
Change the DISPATCH macro to MOD_DISPATCH for consistency with the
other macros.
Add an LKM_ANON #define to eliminate the magic -1 and associated
signed/unsigned warnings.
Add MOD_PRIVATE to support wcd.c's poking around in the lkm structure.
Change source in tree to use the new interface.
Reviewed by: Bruce Evans
by Alan Cox <alc@cs.rice.edu>, and his description of the problem.
The bug was primarily in procfs_mem, but the mistake likely happened
due to the lack of vm system support for the operation. I added
better support for selective marking of page dirty flags so that
vm_map_pageable(wiring) will not cause this problem again.
The code in procfs_mem is now less bogus (but maybe still a little
so.)
implementation #ifdef out. This can be used for now by NFS. As soon
as all the other filesystems' locking is fixed, this can go away.
Print the vnode address in vprint for easier debugging.
1. imgp->image_header needs to be cleared for the bp == NULL && `goto
interpret' case, else exec_fail_dealloc would free it twice after
an error.
2. Moved the vp->v_writecount check in exec_check_permissions() to
near the end. This fixes execve("/dev/null", ...) returning the
bogus errno ETXTBSY. ETXTBSY is still returned for attempts to
exec interpreted files that are open for writing. The man page
is very old and wrong here. It says that ETXTBSY is for pure
procedure (shared text) files that are open for writing or reading.
3. Moved the setuid disabling in exec_check_permissions() to the end.
Cosmetic. It's more natural to dispose of all the error cases
first.
...plus a couple of other cosmetic changes.
Submitted by: bde
