Dave Johnson [Tue, 23 Oct 2007 20:37:22 +0000 (22:37 +0200)]
x86: fix more TSC clock source calibration errors
The previous patch wasn't correctly handling the 'count' variable. If
a CPU gave bad results on the 1st or 2nd run but good results on the
3rd, it wouldn't do the correct thing. No idea if any such CPU
exists, but the patch below handles that case by discarding the bad
runs.
If a bad result (too quick, or too slow) occurs on any of the 3 runs
it will be discarded.
Also updated some comments to explain what's going on.
Signed-off-by: Dave Johnson <djohnson@sw.starentnetworks.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Dave Johnson [Tue, 23 Oct 2007 20:37:22 +0000 (22:37 +0200)]
x86: fix TSC clock source calibration error
I ran into this problem on a system that was unable to obtain NTP sync
because the clock was running very slow (over 10000ppm slow). ntpd had
declared all of its peers 'reject' with 'peer_dist' reason.
On investigation, the tsc_khz variable was significantly incorrect
causing xtime to run slow. After a reboot tsc_khz was correct so I
did a reboot test to see how often the problem occurred:
Test was done on a 2000 Mhz Xeon system. Of 689 reboots, 8 of them
had unacceptable tsc_khz values (>500ppm):
It appears that on rare occasions, mach_countup() is taking longer to
complete than necessary.
I suspect that this is caused by the CPU taking a periodic SMI
interrupt right at the end of the 30ms calibration loop. This would
cause the loop to delay while the SMI BIOS hander runs. The resulting
TSC value is beyond what it actually should be resulting in a higher
tsc_khz.
The below patch makes native_calculate_cpu_khz() take the best
(shortest duration, lowest khz) run of it's 3 calibration loops. If a
SMI goes off causing a bad result (long duration, higher khz) it will
be discarded.
With the patch applied, 300 boots of the same system produce good
results:
Problem was found and tested against 2.6.18. Patch is against 2.6.22.
Signed-off-by: Dave Johnson <djohnson@sw.starentnetworks.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Nick Piggin [Fri, 19 Oct 2007 05:13:02 +0000 (07:13 +0200)]
x86: lock bitops
I missed an obvious one!
x86 CPUs are defined not to reorder stores past earlier loads, so there is
no hardware memory barrier required to implement a release-consistent store
(all stores are, by definition).
So ditch the generic lock bitops, and implement optimised versions for x86,
which removes the mfence from __clear_bit_unlock (which is already a useful
primitive for SLUB).
Signed-off-by: Nick Piggin <npiggin@suse.de> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Linus Torvalds [Tue, 23 Oct 2007 19:04:01 +0000 (12:04 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
9p: v9fs_vfs_rename incorrect clunk order
9p: fix memleak in fs/9p/v9fs.c
9p: add virtio transport
converted many files to use the scatter gather helpers without ensuring
that the necessary headerfile <linux/scatterlist> is included. This
happened to work for ia64, powerpc, sparc64 and x86 because they
happened to drag in that file via their <asm/dma-mapping.h>.
On most of the others this probably broke.
Instead of increasing the header file spider web I choose to include
<linux/scatterlist.h> directly into the affectes files.
Linus Torvalds [Tue, 23 Oct 2007 16:56:11 +0000 (09:56 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
mlx4_core: Increase command timeout for INIT_HCA to 10 seconds
IPoIB/cm: Use common CQ for CM send completions
IB/uverbs: Fix checking of userspace object ownership
IB/mlx4: Sanity check userspace send queue sizes
IPoIB: Rewrite "if (!likely(...))" as "if (unlikely(!(...)))"
IB/ehca: Enable large page MRs by default
IB/ehca: Change meaning of hca_cap_mr_pgsize
IB/ehca: Fix ehca_encode_hwpage_size() and alloc_fmr()
IB/ehca: Fix masking error in {,re}reg_phys_mr()
IB/ehca: Supply QP token for SRQ base QPs
IPoIB: Use round_jiffies() for ah_reap_task
RDMA/cma: Fix deadlock destroying listen requests
RDMA/cma: Add locking around QP accesses
IB/mthca: Avoid alignment traps when writing doorbells
mlx4_core: Kill mlx4_write64_raw()
* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-lguest: (45 commits)
Use "struct boot_params" in example launcher
Loading bzImage directly.
Revert lguest magic and use hook in head.S
Update lguest documentation to reflect the new virtual block device name.
generalize lgread_u32/lgwrite_u32.
Example launcher handle guests not being ready for input
Update example launcher for virtio
Lguest support for Virtio
Remove old lguest I/O infrrasructure.
Remove old lguest bus and drivers.
Virtio helper routines for a descriptor ringbuffer implementation
Module autoprobing support for virtio drivers.
Virtio console driver
Block driver using virtio.
Net driver using virtio
Virtio interface
Boot with virtual == physical to get closer to native Linux.
Allow guest to specify syscall vector to use.
Rename "cr3" to "gpgdir" to avoid x86-specific naming.
Pagetables to use normal kernel types
...
Herbert Xu [Tue, 23 Oct 2007 03:26:25 +0000 (11:26 +0800)]
Fix synchronize_irq races with IRQ handler
As it is some callers of synchronize_irq rely on memory barriers
to provide synchronisation against the IRQ handlers. For example,
the tg3 driver does
tp->irq_sync = 1;
smp_mb();
synchronize_irq();
and then in the IRQ handler:
if (!tp->irq_sync)
netif_rx_schedule(dev, &tp->napi);
Unfortunately memory barriers only work well when they come in
pairs. Because we don't actually have memory barriers on the
IRQ path, the memory barrier before the synchronize_irq() doesn't
actually protect us.
In particular, synchronize_irq() may return followed by the
result of netif_rx_schedule being made visible.
This patch (mostly written by Linus) fixes this by using spin
locks instead of memory barries on the synchronize_irq() path.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Tue, 23 Oct 2007 15:57:35 +0000 (08:57 -0700)]
Merge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
[POWERPC] Add Vitaly Bordug as PPC8xx maintainer
[POWERPC] 4xx: Enable EMAC on Bamboo board
[POWERPC] 4xx: Enable EMAC for PPC405 Walnut board
[POWERPC] 4xx: Fix timebase clock selection on Walnut
[POWERPC] 4xx: Enable EMAC on the PPC 440GP Ebony board
[POWERPC] 4xx: Split early debug output and early boot console for 44x
[POWERPC] 4xx: Enable NEW EMAC support for Sequoia 440EPx.
[POWERPC] 4xx: Add RGMII support for Sequoia 440EPx
Linus Torvalds [Tue, 23 Oct 2007 15:56:50 +0000 (08:56 -0700)]
Merge git://git.infradead.org/mtd-2.6
* git://git.infradead.org/mtd-2.6:
[MTD] [NOR] Fix deadlock in Intel chip driver caused by get_chip recursion
[JFFS2] Fix return value from jffs2_write_end()
[MTD] [OneNAND] Fix wrong free the static address in onenand_sim
[MTD] [NAND] Replace -1 with -EBADMSG in nand error correction code
[RSLIB] BUG() when passing illegal parameters to decode_rs8() or decode_rs16()
[MTD] [NAND] treat any negative return value from correct() as an error
[MTD] [NAND] nandsim: bugfix in initialization
[MTD] Fix typo in Alauda config option help text.
[MTD] [NAND] add s3c2440-specific read_buf/write_buf
[MTD] [OneNAND] onenand-sim: fix kernel-doc and typos
[JFFS2] Tidy up fix for ACL/permissions problem.
Greg Ungerer [Tue, 23 Oct 2007 04:37:54 +0000 (14:37 +1000)]
m68knommu: cleanup m68knommu timer code
Reduce the function pointer mess of the m68knommu timer code by calling
directly to the local hardware's timer setup, and expose the local
common timer interrupt handler to the lower level hardware timer.
Ultimately this will save definitions of all these functions across all
the platform code to setup the function pointers (which for any given
m68knommu CPU family member can be only one set of hardware timer
functions).
Greg Ungerer [Tue, 23 Oct 2007 04:37:54 +0000 (14:37 +1000)]
m68knommu: fix syscall restart handling
Fix system call restart handling. We can call directly to the
restart handler, no need to back track through trap that isn't
even implemented on m68knommu.
Greg Ungerer [Tue, 23 Oct 2007 04:37:54 +0000 (14:37 +1000)]
m68knommu: local module/elf definitions
Up to now m68knommu has been using the asm-m68k/module.h instead of
defining its own. There are recent changes there that we don't need
(fixups specifically). We don't need much support here so it makes
sense to have an m68knommu specific one now.
Jiri Slaby [Tue, 23 Oct 2007 03:45:13 +0000 (20:45 -0700)]
Char: cyclades, fix potential NULL dereference
ztxdone is jumped to even if tty is NULL and tty_wakeup placed after
this label doesn't expect NULLed parameter, so this will cause an oops
in some situations (why they scheduled a wakeup there before remove
bottom half processing patch?).
wakeup only in the case when we have non-null tty struct.
Spotted by Adrian Bunk.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com> Cc: Adrian Bunk <bunk@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Tue, 23 Oct 2007 03:45:12 +0000 (20:45 -0700)]
fix mprotect vma_wants_writenotify prot
Fix mprotect bug in recent commit 3ed75eb8f1cd89565966599c4f77d2edb086d5b0
(setup vma->vm_page_prot by vm_get_page_prot()): the vma_wants_writenotify
case was setting the same prot as when not.
Nothing wrong with the use of protection_map[] in mmap_region(),
but use vm_get_page_prot() there too in the same ~VM_SHARED way.
Signed-off-by: Hugh Dickins <hugh@veritas.com> Cc: Coly Li <coyli@suse.de> Cc: Tony Luck <tony.luck@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Dan Williams [Tue, 23 Oct 2007 03:45:11 +0000 (20:45 -0700)]
md: raid5: fix clearing of biofill operations
ops_complete_biofill() runs outside of spin_lock(&sh->lock) and clears the
'pending' and 'ack' bits. Since the test_and_ack_op() macro only checks
against 'complete' it can get an inconsistent snapshot of pending work.
Move the clearing of these bits to handle_stripe5(), under the lock.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Joel Bertrand <joel.bertrand@systella.fr> Signed-off-by: Neil Brown <neilb@suse.de> Cc: Stable <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
NeilBrown [Tue, 23 Oct 2007 03:45:11 +0000 (20:45 -0700)]
md: fix an unsigned compare to allow creation of bitmaps with v1.0 metadata
As page->index is unsigned, this all becomes an unsigned comparison,
which almost always returns an error.
Signed-off-by: Neil Brown <neilb@suse.de> Cc: Stable <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Alexey Korolev [Mon, 22 Oct 2007 16:55:20 +0000 (17:55 +0100)]
[MTD] [NOR] Fix deadlock in Intel chip driver caused by get_chip recursion
This patch solves kernel deadlock issue seen on JFFF2 simultaneous
operations. Detailed investigation of the issue showed that the kernel
deadlock is caused by tons of recursive get_chip calls.
Signed-off-by: Alexey Korolev <akorolev@infradead.org> Acked-by: Nicolas Pitre <nico@cam.org> Signed-off-by: David Woodhouse <dwmw2@infradead.org>
sctp fails to compile with
net/sctp/sm_make_chunk.c: In function 'sctp_pack_cookie':
net/sctp/sm_make_chunk.c:1516: error: implicit declaration of function 'sg_init_table'
net/sctp/sm_make_chunk.c:1517: error: implicit declaration of function 'sg_set_page'
use the proper include file.
SCTP maintainers Vlad Yasevich and Sridhar Samudrala are CCed.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Implement at32_add_device_cf() which will add a platform_device for
the at32_cf driver (not merged yet). Separate out most of the
at32_add_device_ide() code and use it to implement
at32_add_device_cf() as well.
This changes the API in the following ways:
* The board code must initialize data->cs to the chipselect ID to
use before calling any of these functions.
* The board code must use GPIO_PIN_NONE to indicate unused CF pins.
[AVR32] Implement more at32_add_device_foo() functions
Implement functions for adding platform devices for TWI, MCI, AC97C
and ABDAC. They may need to be modified to cope with platform data,
etc. when the corresponding drivers are ready to be merged, but such
changes are much less likely to conflict than adding support for a
whole new type of device.
This patch adds platform code for PATA devices on the AP7000.
[hskinnemoen@atmel.com: board code left out for now since stk1000
doesn't support IDE out of the box] Signed-off-by: Kristoffer Nyborg Gregertsen <kngregertsen@norway.atmel.com> Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
FUJITA Tomonori [Tue, 23 Oct 2007 07:30:28 +0000 (09:30 +0200)]
parisc: fix sg_page() fallout
arch/parisc/kernel/pci-dma.c: In function 'pa11_dma_map_sg':
arch/parisc/kernel/pci-dma.c:487: error: 'struct scatterlist' has no member named 'page'
arch/parisc/kernel/pci-dma.c: In function 'pa11_dma_unmap_sg':
arch/parisc/kernel/pci-dma.c:508: error: 'struct scatterlist' has no member named 'page'
arch/parisc/kernel/pci-dma.c:508: error: 'struct scatterlist' has no member named 'page'
arch/parisc/kernel/pci-dma.c: In function 'pa11_dma_sync_sg_for_cpu':
arch/parisc/kernel/pci-dma.c:535: error: 'struct scatterlist' has no member named 'page'
arch/parisc/kernel/pci-dma.c:535: error: 'struct scatterlist' has no member named 'page'
arch/parisc/kernel/pci-dma.c: In function 'pa11_dma_sync_sg_for_device':
arch/parisc/kernel/pci-dma.c:545: error: 'struct scatterlist' has no member named 'page'
arch/parisc/kernel/pci-dma.c:545: error: 'struct scatterlist' has no member named 'page'
Heiko Carstens [Tue, 23 Oct 2007 07:28:34 +0000 (09:28 +0200)]
net: fix xfrm build - missing scatterlist.h include
net/xfrm/xfrm_algo.c: In function 'skb_icv_walk':
net/xfrm/xfrm_algo.c:555: error: implicit declaration of function
'sg_set_page'
make[2]: *** [net/xfrm/xfrm_algo.o] Error 1
Cc: David Miller <davem@davemloft.net> Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Tue, 23 Oct 2007 07:49:25 +0000 (09:49 +0200)]
[BLOCK] blk_rq_map_sg: force clear termination bit
Since blk_rq_map_sg() sets the termination bit at the end of the sg
table, we could see it prematurely on the next mapping unless we
force drivers to do a full sg_init_table() prior to each mapping. So
force clear the termination bit to avoid having to put that clear in
the driver for every mapping.
In file included from include/linux/dma-mapping.h:52,
from drivers/base/dma-mapping.c:10:
include/asm/dma-mapping.h: In function 'dma_map_sg':
include/asm/dma-mapping.h:288: error: 'struct scatterlist' has no member named 'page'
include/asm/dma-mapping.h:288: error: 'struct scatterlist' has no member named 'page'
include/asm/dma-mapping.h:288: error: 'struct scatterlist' has no member named 'page'
include/asm/dma-mapping.h:289: error: 'struct scatterlist' has no member named 'page'
include/asm/dma-mapping.h:290: error: 'struct scatterlist' has no member named 'page'
include/asm/dma-mapping.h: In function 'dma_sync_sg_for_cpu':
include/asm/dma-mapping.h:331: error: 'struct scatterlist' has no member named 'page'
drivers/scsi/ps3rom.c: In function 'fetch_to_dev_buffer':
drivers/scsi/ps3rom.c:150: error: 'struct scatterlist' has no member named 'page'
Signed-off-by: Olof Johansson <olof@lixom.net> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Olof Johansson [Tue, 23 Oct 2007 07:12:52 +0000 (09:12 +0200)]
IB/ehca: Fix sg_page() fallout
More fallout from sg_page changes:
drivers/infiniband/hw/ehca/ehca_mrmw.c: In function 'ehca_set_pagebuf_user1':
drivers/infiniband/hw/ehca/ehca_mrmw.c:1779: error: 'struct scatterlist' has no member named 'page'
drivers/infiniband/hw/ehca/ehca_mrmw.c: In function 'ehca_check_kpages_per_ate':
drivers/infiniband/hw/ehca/ehca_mrmw.c:1835: error: 'struct scatterlist' has no member named 'page'
drivers/infiniband/hw/ehca/ehca_mrmw.c: In function 'ehca_set_pagebuf_user2':
drivers/infiniband/hw/ehca/ehca_mrmw.c:1870: error: 'struct scatterlist' has no member named 'page'
Signed-off-by: Olof Johansson <olof@lixom.net> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Takashi Iwai [Mon, 22 Oct 2007 15:20:10 +0000 (17:20 +0200)]
[ALSA] hda-codec - Fix possible array overflow
dac_nids arrays in each codec support code may have up to 5 items
when assigned from the auto-configurator. Some codec codes have
less numbers than the possible max. This patch defines the constant
and fixes the array definitions.
Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Jaroslav Kysela <perex@perex.cz>
The lock grabbed in snd_ctl_empty_read_queue() is hardirq-unsafe but we hold
an hardirq-safe one already, so make the &ctl->read_lock also hard-irq-safe.
Takashi Iwai [Fri, 19 Oct 2007 06:13:40 +0000 (08:13 +0200)]
[ALSA] hda-codec - Fix build without CONFIG_SND_HDA_GENERIC
Fixed the build error from patch_sigmatel.c when built without
CONFIG_SND_HDA_GENERIC by defining a dummy function to return error.
Also, clean up hda_codec.c by removing unneeded ifdefs (the compiler
will optimize out).
Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Takashi Iwai [Thu, 18 Oct 2007 15:33:27 +0000 (17:33 +0200)]
[ALSA] hda-codec - Fix conflict of Master volume in STAC92xx codec
The addition of volume knob as Master volume resulted in conflict with
the existing one by stac92xx_auto_create_hp_ctls().
This patch fixes the conflict, and still keeps the Master control for
codecs without volume knob as much as possible.
Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Trent Piepho [Thu, 18 Oct 2007 08:48:43 +0000 (10:48 +0200)]
[ALSA] snd-bt87x: Make the load_all option work correctly
If the load_all option was turned on all cards would be treated as unknown,
even those which are in the database. Of course, if the card is in the
database there is no reason to use the load_all option. It's there to force
loading when the card isn't in the database. But there are out of date wikis
that say to do this and some distros might turn this option on by default.
So, we keep the load_all option from turning known cards into unknown cards.
Adrian McMenamin [Thu, 18 Oct 2007 08:46:59 +0000 (10:46 +0200)]
[ALSA] protect Dreamcast PCM driver (AICA) from G2 bus effects
The G2 bus on the SEGA Dreamcast connects both the maple peripheral
bus and the AICA sound memory. DMA requests on one can cause the other
to timeout on memory operations.
This patch prevents maple interrupts from causing hiccoughs in the
AICA sound (maple bus code will land in 2.6.24).
There are other cleanups for this (AICA) code - but this is in effect
a regression fix rather than a cleanup.
Signed-off-by: Adrian McMenamin <adrian@mcmen.demon.co.uk> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Rusty Russell [Mon, 22 Oct 2007 01:29:44 +0000 (11:29 +1000)]
Revert lguest magic and use hook in head.S
Version 2.07 of the boot protocol uses 0x23C for the hardware_subarch
field, that for lguest is "1". This allows us to use the standard
boot entry point rather than the "GenuineLguest" string hack.
The standard entry point also clears the BSS and copies the boot parameters
and commandline for us, saving more code.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Mon, 22 Oct 2007 01:24:24 +0000 (11:24 +1000)]
generalize lgread_u32/lgwrite_u32.
Jes complains that page table code still uses lgread_u32 even though
it now uses general kernel pte types. The best thing to do is to
generalize lgread_u32 and lgwrite_u32.
This means we lose the efficiency of getuser(). We could potentially
regain it if we used __copy_from_user instead of copy_from_user, but
I'm not certain that our range check is equivalent to access_ok() on
all platforms.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Acked-by: Jes Sorensen <jes@sgi.com>
Rusty Russell [Mon, 22 Oct 2007 01:24:23 +0000 (11:24 +1000)]
Example launcher handle guests not being ready for input
We currently discard console and network input when the guest has no
input buffers. This patch changes that, so that we simply stop
listening to that fd until the guest refills its input buffers.
This is particularly important because hvc_console without interrupts
does backoff polling and so often lose characters if we discard.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Mon, 22 Oct 2007 01:24:22 +0000 (11:24 +1000)]
Update example launcher for virtio
Implements virtio-based console, network and block servers. The block
server uses a thread so it's async, which is an improvement over the
old synchronous implementation (but a little more complex).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Mon, 22 Oct 2007 01:24:21 +0000 (11:24 +1000)]
Lguest support for Virtio
This makes lguest able to use the virtio devices.
We change the device descriptor page from a simple array to a variable
length "type, config_len, status, config data..." format, and
implement virtio_config_ops to read from that config data.
We use the virtio ring implementation for an efficient Guest <-> Host
virtqueue mechanism, and the new LHCALL_NOTIFY hypercall to kick the
host when it changes.
We also use LHCALL_NOTIFY on kernel addresses for very very early
console output. We could have another hypercall, but this hack works
quite well.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Mon, 22 Oct 2007 01:03:40 +0000 (11:03 +1000)]
Virtio helper routines for a descriptor ringbuffer implementation
These helper routines supply most of the virtqueue_ops for hypervisors
which want to use a ring for virtio. Unlike the previous lguest
implementation:
1) The rings are variable sized (2^n-1 elements).
2) They have an unfortunate limit of 65535 bytes per sg element.
3) The page numbers are always 64 bit (PAE anyone?)
4) They no longer place used[] on a separate page, just a separate
cacheline.
5) We do a modulo on a variable. We could be tricky if we cared.
6) Interrupts and notifies are suppressed using flags within the rings.
Users need only get the ring pages and provide a notify hook (KVM
wants the guest to allocate the rings, lguest does it sanely).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Cc: Dor Laor <dor.laor@qumranet.com>
Rusty Russell [Mon, 22 Oct 2007 01:03:39 +0000 (11:03 +1000)]
Virtio console driver
This is an hvc-based virtio console driver. It's suboptimal becuase
hvc expects to have raw access to interrupts and virtio doesn't assume
that, so it currently polls.
There are two solutions: expose hvc's "kick" interface, or wean off hvc.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Mon, 22 Oct 2007 01:03:38 +0000 (11:03 +1000)]
Block driver using virtio.
The block driver uses scatter-gather lists with sg[0] being the
request information (struct virtio_blk_outhdr) with the type, sector
and inbuf id. The next N sg entries are the bio itself, then the last
sg is the status byte. Whether the N entries are in or out depends on
whether it's a read or a write.
We accept the normal (SCSI) ioctls: they get handed through to the other
side which can then handle it or reply that it's unsupported. It's
not clear that this actually works in general, since I don't know
if blk_pc_request() requests have an accurate rq_data_dir().
Although we try to reply -ENOTTY on unsupported commands, ioctl(fd,
CDROMEJECT) returns success to userspace. This needs a separate
patch.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Cc: Jens Axboe <jens.axboe@oracle.com>
Rusty Russell [Mon, 22 Oct 2007 01:03:37 +0000 (11:03 +1000)]
Net driver using virtio
The network driver uses two virtqueues: one for input packets and one
for output packets. This has nice locking properties (ie. we don't do
any for recv vs send).
TODO:
1) Big packets.
2) Multi-client devices (maybe separate driver?).
3) Resolve freeing of old xmit skbs (Christian Borntraeger)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: netdev@vger.kernel.org
Rusty Russell [Mon, 22 Oct 2007 01:03:36 +0000 (11:03 +1000)]
Virtio interface
This attempts to implement a "virtual I/O" layer which should allow
common drivers to be efficiently used across most virtual I/O
mechanisms. It will no-doubt need further enhancement.
The virtio drivers add buffers to virtio queues; as the buffers are consumed
the driver "interrupt" callbacks are invoked.
There is also a generic implementation of config space which drivers can query
to get setup information from the host.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Cc: Dor Laor <dor.laor@qumranet.com> Cc: Arnd Bergmann <arnd@arndb.de>
Rusty Russell [Mon, 22 Oct 2007 01:03:36 +0000 (11:03 +1000)]
Boot with virtual == physical to get closer to native Linux.
1) This allows us to get alot closer to booting bzImages.
2) It means we don't have to know page_offset.
3) The Guest needs to modify the boot pagetables to create the
PAGE_OFFSET mapping before jumping to C code.
4) guest_pa() walks the page tables rather than using page_offset.
5) We don't use page_offset to figure out whether to emulate: it was
always kinda quesationable, and won't work for instructions done
before remapping (bzImage unpacking in particular).
6) We still want the kernel address for tlb flushing: have the initial
hypercall give us that, too.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Mon, 22 Oct 2007 01:03:35 +0000 (11:03 +1000)]
Allow guest to specify syscall vector to use.
(Based on Ron Minnich's LGUEST_PLAN9_SYSCALL patch).
This patch allows Guests to specify what system call vector they want,
and we try to reserve it. We only allow one non-Linux system call
vector, to try to avoid DoS on the Host.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
projects / linux-2.6 / log