the 'debugging' section of any HEAD kernel and enable for the mainstream
ones, excluding the embedded architectures.
It may, of course, enabled on a case-by-case basis.
Sponsored by: Sandvine Incorporated
Requested by: emaste
Discussed with: kib
have code that detects this and makes two
transmit descriptors. However its possible
that the algorithm detects when the second
page is not used (when the data aligns perfectly
to the bottom of the page). This caused a 0
len descriptor to be added which locks up the
rge device. Skip such things with a continue.
JC provided this patch... Thanks JC :-)
Obtained from: JC (c.jayachandran@gmail.com)
Specifically on an SMP kernel it was observed that if both the
processors are doing an exit1() via ast()->postsig()->sigexit()
then we will deadlock.
This happens because exit1() calls vmspace_exit() that in turn
calls pmap_invalidate_all(). This function tries to do a
smp_rendezvous() which blocks because the other processor is not
responding to IPIs - because it too is doing AST processing with
interrupts disabled.
The platform that supports SMP currently is a SWARM with a dual-core Sibyte
processor. The kernel config file to use is SWARM_SMP.
Reviewed by: imp, rrs
disabling interrupts.
Simplify register usage - we can directly load 'curpcb' into 'k1' after
interrupts are disabled. There is no need to do so indirectly through 'a1'.
The only reason we need to have the sb_load64() and sb_store64()
functions in assembly is to cheat the compiler and generate the
'ld' and 'sd' instructions which it otherwise will not do when
compiling for a 32-bit architecture. There are some 64-bit
registers in the SCD unit that must be accessed using 64-bit
load and store instructions.
The basic idea is to use a the same virtual address as a window onto
distinct physical memory locations - one per processor. The physical
address that you access through this mapping depends on which cpu you
are currently executing on. We can now use the same virtual address
on any processor to access its per-cpu area.
The details are:
- The virtual address for 'struct pcpu *pcpup' is obtained by
stealing 2 pages worth of KVA in pmap_bootstrap().
- The mapping from the constant virtual address to a distinct
physical page is done in cpu_pcpu_init() through a wired TLB entry.
- A side-effect of this is that we reserve 2 pages worth of memory
for the pcpu but in reality it needs much less than that. The unused
memory is now used as the boot stack for the BSP and APs.
Remove SMP-specific bits from locore.S. The plan is to use a separate
mpboot.S for AP bootstrap.
Discussed on: freebsd-mips
Approved by: imp (mentor)
interrupts must be disabled through the
page_zero's or copys etc. Note that the
temporary mapping used by panic's may
cause us pain since int's may not be disabled.
When we get dumps working we may have to revist
this. Note that with this fix the build got
much much further.. until it hung on disk IO (I
would imagine thats the rge/msgring driver acting
up).
above 4Gig. If so when we add the base address with
the size we will wrap. So for now we just ignore
such memory and only use what we can. When we
get 64 bit working then we will be much better ;->
lacking a copyright/license statement. All these files were in the
Cavium FreeBSD source drop and appear to be written by Cavium (some
are nearly verbatim copies of files from the cnusers' 1.9.0 SDK, which
also uses this copyright).
routine in one place was setting the valid2 bit to
2 not 1. This meant the PTE was NOT valid and so
you would crash.
In Zero Page there was a incorrect setting of
the valid bit AFTER the actual zero (opps)..
Hopefully this will fix the 0xc0000000 crashes
that I have been seeing (unless of course there are
other problems with these old hacks of mine to get
to memory above 512Meg)
the RMI/XLR has the COP0 and COP2 bits enabled
Plus it needs SX too. Thanks again for JC in
catching this ;-)
Submitted by: JC (jayachandranc@netlogicmicro.com
to invalidate memory chunk that starts or ends in the middle of
cache line.
This was responsible for one half of the problem preventing umass
to work reliably on some MIPS32 platforms. USBng needs to stop
sharing cache lines between DMA-able memory and other structures
to cure the other half.
Discussed with: imp, gonzo
locks up in make buildworld.
You need to follow the mips wiki for building
the nfs partition and setup things to mount there
(in the conf and in your bootp setup).
This is a workaround for the fact that the CFE is compiled as a 64-bit
application and therefore sets the SR_KX bit every time we call into
it (for e.g. console).
A TLB miss for any address above 0xc0000000 with the SR_KX bit set will
end up at the XTLB exception vector. We workaround this by copying the
standard TLB handler at the XTLB exception vector.
Approved by: imp (mentor)
through fork_trampoline.
This was caused because we were clearing the SR_INT_IE and setting
SR_EXL bits of the status register at the same time. This meant
that if an interrupt happened while this MTC0 was making its way
through the pipeline the exception processing would see the
status register with SR_EXL bit set. This in turn would mean that
the COP_0_EXC_PC would not be updated so the return from exception
would be to an incorrect address.
It is easy to verify this fix by a program that forks in a loop
and the child just exits:
while (1) {
pid_t pid = vfork();
if (pid == 0)
_exit(0);
if (pid != -1)
waitpid(pid, NULL, 0);
}
Also remove two instances where we set SR_EXL bit gratuitously in exception.S.
Approved by: imp (mentor)
obio. Take advantage of the fact that obio only really supports uart
at the moment to use the uart bus tag always for IOPORT allocations.
# this needs to be redone to conform to FreeBSD standards and allow for
# additional drivers for SoC hardware to attach
in the pcb at the td_lock offset, rather than in the struct thread at
the td_lock offset. And we were storing a pointer to the old thread
rather than to the mutex. Why this didn't always kill us, I'll never
know.
Fix an obsolete comment and update the prototype in the comments.
Also note what variables are in what registers since this function is
a little complex...
neel@ found this problem and proposed this fix. This cures a number
of different problem reports out there, and gets us booting octeon to
the login prompt...
Submitted by: neel@
Reviewed by: rrs@, gonzo@
symbols resolving in DDB
- When zeroing .bss/.sbss do not round end address to page boundary,
it's not neccessary and might destroy data pased by trampoline or
boot loader
available for use. All data below this address considered to be used
by kernel. Along with kernel own data it might be symbol tables
prepeared by trampoline code, boot loader service data passed for
further analysis by kernel, etc... By default kernel_kseg0_end points
to the end of loaded kernel.
- Introduce mips_postboot_fixup function. It checks for symbol information
copied by ELF trampoline and passes it to KDB
