src/Makefile.inc1 rev. 1.590, it can allow installing a world
cross-built for a different arch over the live system. The procedure
is more or less as follows:
cp -R /rescue /rescue.old
make installkernel TARGET_ARCH=foo
make -DNO_RTLD installworld TARGET_ARCH=foo
^^^^^^^^^
PATH=/rescue.old
chflags noschg /libexec/ld-elf.so.1
cp /usr/obj/foo/usr/src/libexec/rtld/ld-elf.so.1 /libexec
chflags schg /libexec/ld-elf.so.1
<ditto for ld-elf32.so.1 if installing for amd64>
reboot
I.e., not only copy them to a scratch dir, but also make them use saved
copies of libraries and locale files. That gives us several benefits:
1) ABI breakages should no longer affect installworld over the live system.
2) It becomes safe to run installworld while still running the old kernel.
However, it can be reasonable to save the old /rescue before that to be
able to run the old reboot(8), as the new binaries are rather likely to
fail with the old kernel. Anyhow, it's now possible to upgrade a system
in a single reboot _reliably_.
3) With a bit of hackery around rtld(8), it becomes possible to do destructive
cross-installs, e.g., i386->amd64 over the live system.
The only shared item left between the old and new systems is rtld(8),
which cannot be run from a saved copy easily because its full
pathname is stored in the respective field of each ELF executable.
(In theory, that field could be overridden, e.g., from the environment,
but this can lead to security issues.) That's why a destructive
cross-install isn't possible w/o hackery yet.
Fruitful ideas by: ru
Reviewed by: ru
Tested with: audit(4)
to change the freq before the other CPUs are active. The current code
always attempts to change all CPUs to match each other, and the requisite
sched_bind() call won't work before APs are launched.
/dev/agpgart and agp_free_res() frees resources like the BAR for the
aperture. Splitting this up lets chipset-specific detach routines
manipulate the aperture during their detach routines without panicing.
MFC after: 1 week
Reviewed by: anholt
* Do not hold any locks over calls to copyin/copyout.
* Clean up some #ifdefs
* fix a possible mbuf leak when NAT fails on policy routed packets
PR: 117216
- Select a tag gains ability to optionally save new tags
off in the timewait system.
- When looking up associations do not give back a stcb that
is in the about-to-be-freed state, and instead continue
looking for other candiates.
- New function to query to see if value is in time-wait.
- Timewait had a time comparison error that caused very
few vtags to actually stay in time-wait.
- When setting tags in time-wait, we now use the time
requested NOT a fixed constant value.
- sstat now gets the proper associd when we do the query.
- When we process an association, we expect the tag chosen
(if we have one from a cookie) to be in time-wait. Before
we would NOT allow the assoc up by checking if its good.
In theory this should have caused almost all assoc not
to come up except for the time-comparison bug above (this
bug was hidden by the time comparison bug :-D).
- Don't save tags for nonce values in the time-wait cache
since these are used only during cookie collisions and do
not matter if they are unique or not.
MFC after: 1 week
to tune pthread mutex performance:
1. LIBPTHREAD_SPINLOOPS
If a pthread mutex is being locked by another thread, this environment
variable sets total number of spin loops before the current thread
sleeps in kernel, this saves a syscall overhead if the mutex will be
unlocked very soon (well written application code).
2. LIBPTHREAD_YIELDLOOPS
If a pthread mutex is being locked by other threads, this environment
variable sets total number of sched_yield() loops before the currrent
thread sleeps in kernel. if a pthread mutex is locked, the current thread
gives up cpu, but will not sleep in kernel, this means, current thread
does not set contention bit in mutex, but let lock owner to run again
if the owner is on kernel's run queue, and when lock owner unlocks the
mutex, it does not need to enter kernel and do lots of work to resume
mutex waiters, in some cases, this saves lots of syscall overheads for
mutex owner.
In my practice, sometimes LIBPTHREAD_YIELDLOOPS can massively improve performance
than LIBPTHREAD_SPINLOOPS, this depends on application. These two environments
are global to all pthread mutex, there is no interface to set them for each
pthread mutex, the default values are zero, this means spinning is turned off
by default.
caused a segfault. It turns out that in pre-7.0 systems if you do
getenv("amd_enable=YES") it will return the setting of the environment
variable "amd_enable" but now it returns NULL. I think I found the
places where sysinstall was potentially relying on that old behavior.
Fix is to make a copy of the string to be used for the getenv(3) call,
look for a '=' character in it, and replace it with '\0' if one is
found. Stuck to sysinstall's typical coding standards despite urges
to do otherwise.
PR: 117642
MFC after: 2 days
