| Age | Commit message (Collapse) | Author |
|---|
|
from a USB serial number, as recommended by the umass spec.
ok dlg@
|
|
until they're zombies and then send them signals (for intr mounts). Until
that is untangled, the sigacts change is unsafe. sthen@ was the victim
for this one
|
|
The problems during the hackathon were not caused by this (most likely).
prodded by deraadt@ and beck@
|
|
this will treat vndX the same as svndX.
ok and gcc2 fixes from deraadt@
|
|
when leaving. when you're handling an interrupt it is masked.
whacking the chip is work for no gain.
modify the interrupt handler so it only processes the rings once
rather than looping over them until it runs out of work to do
looping in the isr is bad for several reasons:
firstly, the chip does interrupt mitigation so you have a
decent/predictable amount of work to do in the isr. your first loop
will do that chunk of work (ie, it pulls off 50ish packets), and
then the successive looping aggressively pull one or two packets
off the rx ring. these extra loops work against the benefit that
interrupt mitigation provides.
bus space reads are slow. we should avoid doing them where possible
(but we should always do them when necessary).
doing the loop 5 times per isr works against the mclgeti semantics.
it knows a nic is busy and therefore needs more rx descriptors by
watching to see when the nic uses all of its descriptors between
interrupts. if we're aggressively pulling packets off by looping
in the isr then we're skewing this check.
ok deraadt@
|
|
eject events, then the ppb's interrupt handler can be run at IPL_BIO.
This makes a huge difference on modern server with mpi/bge/bnx.
In snapshots for about 2 weeks now...
|
|
this probably means the controller is dead -- as in, a hot plug eject
event. in that case, mark the driver dead and try harder to return -1
in the interrupt handler.
tested by many
ok dlg, miod did not object
|
|
|
|
attaching since theyre useless without mpath.
the path drivers ask mpath if its ok to use the device before doing their
own matches (this is so mpath can prevent paths attaching to itself), so
im just adding this check there.
this uses code from miod to walk cfdata for the mpath entry and then checks
its state. this is ok because mpath is only attached in one place, so there
arent multiple cfdata entries for it.
ok krw@ deraadt@ miod@ matthew@
|
|
test it now that it's in-tree. ok deraadt ("It causes no harm")
|
|
addresses from process context by scheduling a work queue.
Fixes an assertwaitok panic. ok henning
|
|
ok deraadt
|
|
off anymore.
ok miod
|
|
ok miod
|
|
(pretty easy, since the bootdv is always wd0)
|
|
hubs are known during autoconf. Then, pick the most populated 2GB window
as our DMA memory window. xbridge(4) can thus program the correct settings
regardless of the order in which the xbow(4) attach.
|
|
chance to interrupt the boot procedure before booting the default kernel,
just like (most) of our other hardware platforms. Interactive boots, where
you interrupt the firmware boot procedure and answer "Y" to the
"Interact with IPL?" question will now wait for human intervention at the
boot> prompt (without a timeout). Should reduce frustration when trying to
boot an alternative kernel.
ok deraadt@, miod@, jsing@, todd@
|
|
evil problems with non-equivalent aliases.
|
|
evil problems with non-equivalent aliases.
|
|
matter, since we shouldn't be using the FPU inside the kernel, but we need to
initialize %cr30 with something, and this is more consistent than using the
address of the pcb.
|
|
|
|
|
|
what the previous IO was. Less chance of copy and paste errors.
Suggested by miod@.
|
|
|
|
|
|
(interrupt was not for me), 1 (positive interrupt was for me), or -1
(i am not sure...). We have continued with this practice in as many
drivers as possible, throughout the tree.
This makes some of the architectures use that information in their
interrupt handler calling code -- if 1 is returned (and we know
this specific machine does not have edge-shared interrupts), we
finish servicing other possible handlers on the same pin. If the
interrupt pin remains asserted (from a different device), we will
end up back in the interrupt servicing code of course... but this is
cheaper than calling all the chained interrupts on a pin.
This does of course count on shared level interrupts being properly
sorted by IPL.
There have been some concerns about starvation of drivers which
incorrectly return 1. Those drivers should be hunted down so that
they return -1.
ok and help from various people. In snaps for about a week now.
|
|
|
|
page so that um_anfree will free it for us.
uvm_anfree does a pmap_page_protect(, VM_PROT_NONE) just before it frees the
page, so we don't need to do it here ourselves.
ok ariane@
|
|
(uvm_atopg) and use it in uvm_km_doputpage to replace some handrolled
code. Shrinks the kernel a trivial amount.
ok beck@ and miod@ (who suggested i name it uvm_atopg not uvm_atop)
|
|
At various times diffs have had debugging that checked that we don't
insert a page into the tree on top of an existing page, leaking that
page's references. Until the recent hackathon (and introduction if
uvm_pagealloc_multi) the bufcache for example did a rb tree look up on
insert to check (under #ifdef DEBUG || 1) so instead just check it on
pageinsert every time, since RB_INSERT returns any duplicates so this
check is pretty much free.
``emphatically yes'' beck@
|
|
global hash I forgot to remove the has declarations from struct uvm. So
remove them now.
pointed out by blambert@, ok beck@
|
|
argument to wdcreset(), to have it skip waiting until active channels see
their BUSY bit clear in the status register.
Use this feature in the resume path, during the first reset operation. The
first reset is supposed to only wake up the controller, and the disks don't
come back until the second reset is issued, therefore waiting for them to
report themselves as ready after the first reset, but before the second, is
moot - and as a matter of fact some controllers, such as the AMD 754 and
clones/offspring (e.g. Geode) keep the BUSY bit asserted after the first
reset.
Last, but not least, make sure wd@ata invokes wd_get_params() again before
returning from the resume code, as we will still be using polled transfers
for a short while.
This causes the Lemote Yeelong to resume within less than one second, instead
of the lousy 30 seconds wait between the two resets; and the wd_get_params()
voodoo prevents it from getting spurious ide interrupts afterwards.
wd_get_params() magic from dlg; rest of the work by yours truly after enough
prodding by dlg@ and pirofti@, among others. ok deraadt@ dlg@
|
|
(interrupt was not for me), 1 (positive interrupt was for me), or -1
(i am not sure...). We have continued with this practice in as many
drivers as possible, throughout the tree.
This makes some of the architectures use that information in their
interrupt handler calling code -- if 1 is returned (and we know
this specific machine does not have edge-shared interrupts), we
finish servicing other possible handlers on the same pin. If the
interrupt pin remains asserted (from a different device), we will
end up back in the interrupt servicing code of course... but this is
cheaper than calling all the chained interrupts on a pin.
This does of course count on shared level interrupts being properly
sorted by IPL.
There have been some concerns about starvation of drivers which
incorrectly return 1. Those drivers should be hunted down so that
they return -1.
ok drahn
|
|
(interrupt was not for me), 1 (positive interrupt was for me), or -1
(i am not sure...). We have continued with this practice in as many
drivers as possible, throughout the tree.
This makes some of the architectures use that information in their
interrupt handler calling code -- if 1 is returned (and we know
this specific machine does not have edge-shared interrupts), we
finish servicing other possible handlers on the same pin. If the
interrupt pin remains asserted (from a different device), we will
end up back in the interrupt servicing code of course... but this is
cheaper than calling all the chained interrupts on a pin.
This does of course count on shared level interrupts being properly
sorted by IPL.
There have been some concerns about starvation of drivers which
incorrectly return 1. Those drivers should be hunted down so that
they return -1.
(other architectures will follow)
ok kettenis drahn dlg miod
|
|
to handle that with the sigacts change.
problem observed and diagnosed by claudio@ and kettenis@
|
|
year.
acpi needs to use the apm definitions so that apmd speaks the same
language as it, so it uses the ones in apmvar.h these days.
``sure'' marco@
|
|
Found by LLVM/Clang Static Analyzer.
ok guenther@
|
|
Found by LLVM/Clang Static Analyzer.
ok claudio@
|
|
Found by LLVM/Clang Static Analyzer.
ok henning@ krw@ claudio@
|
|
ok pirofti@ guenther@
|
|
are past. Use CLR() and SET() to modify necessary flags while leaving
the flags used by the buffer cache in peace.
Should make bufcache code much less confused about the state of the
bufs used in reading/writing disklabels. Other such flag abuses no
doubt await a visit.
Errors in original diff found by miod@.
ok beck@ deraadt@
|
|
and cause build failure. Bad guenther@ no cookies.
commitski! mikeb@
|
|
be shared (p_sigignore, p_sigcatch, P_NOCLDSTOP, P_NOCLDWAIT) moves
to struct sigacts, wihle stuff that should be per rthread (ps_oldmask,
SAS_OLDMASK, ps_sigstk) moves to struct proc. Treat the coredumping
state bits (ps_sig, ps_code, ps_type, ps_sigval) as per-rthread
until our locking around coredumping is better.
Oh, and remove the old SunOS-compat ps_usertramp member.
"I like the sound of this" tedu@
|
|
|
|
using the -MD option to cc, with -MP, -MT, and -MF where needed, converting
"make depend" to a no-op. This increases parallelism for those using "make -j"
and keeps the dependencies up to date with each compilation automatically.
sparc and vax users will need to rebuild gcc with support for the
-M[PTF] options before config'ing with this diff.
|
|
|
|
|
|
|
|
ok claudio@
|
|
RTL8103E, from FreeBSD.
ok sthen@
|
