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
struct direct, not using UFS' definition of DIRBLKSIZ, using directory
seek cookies to make reading non-UFS directories reliable
(e.g. cd9660, ext2fs).
A special thanks to Robert Eckardt for providing an ISC binary of GNU
ls so that I could test these changes.
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.)
selection loop was merged with the poll_for_work loop. We cannot assume
that the SCB for the selection timeout is the current SCB. Instead we
must look at the SCB at the head of the waiting for selection list.
This fixes part of a problem reported by David Malone, but does not explain
why he was getting selection timeouts in the first place.
the directory format (ext2fs, cd9660). For these filesystems, it must use
cookies to find the correct offset to use for subsequent reads. Without it,
linux /bin/ls tends to loop re-reading the same block over and over again.
2.2 candidate.
(see LINT). There is a new low-level console type that is more suitable
for use with gdb-remote.
Fixed setting of speed at probe time for the serial console (if any).
Reviewed by: dfr
Bump the timeout for an "ordered tag" recovery action from 1 to 5 seconds.
Remove the multiple timeout panic. Its very easy to get into a situation
where a timedout command will time out a second time even though the
recovery code is working fine. A good example is:
1) Command times out during recovery
2) reset the timeout for the command
3) Recovery actions complete and all transactions are requeued
4) second timeout fires off which puts us back into recovery bogusly
5) another transaction that timedout once during the first recovery action
times out causing the panic.
In essence, the correct solution to the problem is to put every transaction
back up into the work queue and have their timeout handling done in the same
way that all commands are handled. The CAM layer makes this easy, so it
will have to wait until then.
either by looking it up in the array of pending, per target, untagged
transactions, or by using the tag value passed in during the identify. The
old code only direct indexed for tagged transactions. This makes the
"findSCB" routine only necessary when SCB paging is enabled, so appropriately
conditionalize it. This greatly simplifies the non SCB paging code flow.
if all registers are 0xff.
This allows me to run with flags 0xc0ff on my IBM-DMCA-21440 disk, which
gives 5MB/sec sequential read :-)
If you have a laptop, try adding flag 0x4000 to your disk, and tell me if
it makes any difference for you.
centric rather than VM-centric to fix a problem with errors not being
detectable when the header is read.
Killed exech_map as a result of these changes.
There appears to be no performance difference with this change.
Lookup isn't done every time the system goes idle now, but it can still
take > 1800 instructions in the worst case, so if cpu interrupts are kept
disabled then it might lose 20 characters of sio input at 115200 bps.
Fixed style in vm_page_zero_idle().
functions if DDB is available. The remaining occurences are usually
only inlined and thus not available in DDB.
I'm sure Bruce will have 23 additions to these 30 lines of code, but
at least it's a starting point. ;-)
change typematic rate, or the X server (XFree86 or Accelerated X)
starts up.
So far, there have been two independent reports from Dell Latitude XPi
notebook/laptop owners. The Latitude seems to be the only system which
suffers from this problem. (I don't know the problem is with the
entire Latitude line or with only some Latitude models) No problem
report has been heard about other systems (I certainly cannot
reproduce the problem in my -current and 2.2 systems).
In 3.0-CURRENT, 2.2-RELEASE and 2.2-GAMMA-970310, when programming the
keyboard LED/repeat-rate, `set_keyboard()' in `syscons' tells the
keyboard controller not to generate keyboard interrupt (IRQ1) and then
enable tty interrupts, expecting the keyboard interrupt doesn't occur.
It appears that somehow Latitude's keyboard controller still generates
the keyboard interrupt thereafter, and `set_keyboard()' doesn't see
the return code from the keyboard because it is consumed by the
keyboard interrupt handler.
The patch entirely disables tty interrupts while setting LED and
typematic rate in `set_keyboard()', making the routine behave more
like the previous versions of `syscons' (versions in 2.1.X and
2.2-ALPHA, -BETA, and some -GAMMAs). The reporter said this patch
eliminated the problem.
(I also found another typo/bug, but the reporter and I found that it
wasn't the cause of the problem...)
This should go into RELENG_2_2.
