read requests to its consumer. It has been developed to address
the problem of a horrible read performance of a 64k blocksize FS
residing on a RAID3 array with 8 data components, where a single
disk component would only get 8k read requests, thus effectively
killing disk performance under high load. Documentation will be
provided later. I'd like to thank Vsevolod Lobko for his bright
ideas, and Pawel Jakub Dawidek for helping me fix the nasty bug.
calls are not used by libthr in RELENG_6 and HEAD, it is only used by
the libthr in RELENG-5, the _umtx_op system call can do more incremental
dirty works than these two system calls without having to introduce new
system calls or throw away old system calls when things are going on.
unsuspecting users.
- Add a comment in NOTES about experimental status of SCHED_ULE.
- Make warning about experimental status in sched_ule(4) a bit
stronger.
Suggested and reviewed by: dougb
Discussed on: developers
MFC after: 3 days
Reviewed by: jmg and jb
Forced commit to provide a more meaningful commit message.
John Michael-Gurney is responsible for fixing all the vast majority of the
issues that arose with PCI support.
Kris Kennaway and John Birrell pushed heavily on stability issues. Kris is to
thank for it having even reached its current level of stability. When it
reaches a production state, we will have John to thank for driving it forward.
The following items are needed before sun4v can be used from CVS:
- binutils/gas/config/tc-sparc.c needs to be pulled in from a modern
version of binutils.
- an updated version of the loader needs to be pulled in from perforce
The following items need to be addressed before sun4v can be considered for
deployment:
- The T1 has a fairly minimal floating point unit, fsqrt and a number of
other opcodes are not implemented. Emulation for unimplemented opcodes
needs to be added.
- Running against local disk a race condition exists whereby a cpu will
sometimes run off into the weeds at tl > 2.
- The MD code for interrupt handling will crash on the first ATA interrupt.
- The BCM chipset (bge) is not set up correctly on the T1000. The current
workaround is to use an add in Intel GigE card (em).
- With KSE in the kernel signals are often not directed to the right
process / thread within a process. This causes threaded processes
to frequently become unkillable and gdb to not be reliable when debugging
a threaded process. This issue is slowly being resolved on -threads.
Many thanks go to Sun for having lent / donated a T2000 to both me and
John Birrell.
or not the OS has to wait for RX_RDY or TX_RDY to be set before the OS sets
the control code in the control/status register. Looking at the interface
design, it seems that RX_RDY and TX_RDY are probably there to protect
access to the data register and have nothing to do with the control/status
register. Nevertheless, try to take what I think is the more conservative
approach and always wait for the appropriate [TR]X_RDY flag to be set
before writing any of the WR_NEXT, WR_END, RD_START, or RD_NEXT control
codes to the control/status register.
commits. For some reason I thought the scale factor was a shift count
rather than the multiplicand (that is, I thought leal (%eax,%edx,4) was
going to generate %eax + %edx << 4 rather than %eax + %edx * 4). What
I need is to multiply by 16 to convert a real-mode (seg, offset) tuple
into a flat address. However, the max multiplicand for scaled/index
addressing on i386 is 8, so go back to using a shl and an add.
- Convert two more inter-register mov instructions where we don't need to
preserve the source register to xchg instructions to keep our space
savings.
Tested by: Ian FREISLICH if at hetzner.co.za
MFC after: 1 week
This allows the user to specify that the root filoesystem should be
an MFS or teh actual medium itself.
Also a new command syntax, along with an interactive mode by default,
with crossinvocation memory of the last used values as defaults for the
current invocation.
Submitted by:jmeloatfreebsdbrasil,com-br
Submitted by:
Reviewed by:
Approved by:
Obtained from:
MFC after:
Security:
Move the relocation definitions to the common elf header so that DTrace
can use them on one architecture targeted to a different one.
Add the additional ELF types defines in Sun's "Linker and Libraries"
manual.
address, to avoid confusing the users that a full address is
always required.
Submitted by: Josh Paetzel <josh@tcbug.org> (through freebsd-doc)
MFC after: 3 days
partition size. On 32-bit platforms sizeof(long) < sizeof(off_t)
and using strtol(3) would prevent partitions larger than 4G
sectors or beyond 4G blocks.
PR: bin/103991
MFC after: 3 days
there are enough places in the DTrace kernel/module sources that
having a header that gathers together all the individual elf headers
is convenient.
Note that the Solaris compatibility definions are conditionally
included iff _SOLARIS_C_SOURCE is defined.
by Sun's CDDL and this file is only intended for inclusion where
_SOLARIS_C_SOURCE is defined (with the assumption that the code
being compiled is licensed under the CDDL too).
restore it directly and skip chmod() during the post-extract fixup.
In particular, bsdtar -xm now completely skips the post-extract fixup
for directories, which produces a noticable speedup in that case.
unmount when mp structure is reused while waiting for coveredvp lock.
Introduce struct mount generation count, increment it on each reuse and
compare the generations before and after obtaining the coveredvp lock.
Reviewed by: tegge, pjd
Approved by: pjd (mentor)
MFC after: 2 weeks
