mckusick [Wed, 28 Mar 2012 21:34:55 +0000 (21:34 +0000)]
MFC of 232351, 233438, and 233629
MFC reviewed by: kib
MFC 232351:
This change avoids a kernel deadlock on "snaplk" when using
snapshots on UFS filesystems running with journaled soft updates.
This is the first of several bugs that need to be fixed before
removing the restriction added in -r230250 to prevent the use
of snapshots on filesystems running with journaled soft updates.
The deadlock occurs when holding the snapshot lock (snaplk)
and then trying to flush an inode via ffs_update(). We become
blocked by another process trying to flush a different inode
contained in the same inode block that we need. It holds the
inode block for which we are waiting locked. When it tries to
write the inode block, it gets blocked waiting for the our
snaplk when it calls ffs_copyonwrite() to see if the inode
block needs to be copied in our snapshot.
The most obvious place that this deadlock arises is in the
ffs_copyonwrite() routine when it updates critical metadata
in a snapshot and tries to write it out before proceeding.
The fix here is to write the data and indirect block pointer
for the snapshot, but to skip the call to ffs_update() to
write the snapshot inode. To ensure that we will never have
to update a pointer in the inode itself, the ffs_snapshot()
routine that creates the snapshot has to ensure that all the
direct blocks are allocated as part of the creation of the
snapshot.
A less obvious place that this deadlock occurs is when we hold
the snaplk because we are deleting a snapshot. In the course of
doing the deletion, we need to allocate various soft update
dependency structures and allocate some journal space. If we
hit a resource limit while doing this we decrease the resources
in use by flushing out an existing dirty file to get it to give
up the soft dependency resources that it holds. The flush can
cause an ffs_update() to be done on the inode for the file that
we have selected to flush resulting in the same deadlock as
described above when the inode that we have chosen to flush
resides in the same inode block as the snapshot inode that we hold.
The fix is to defer cleaning up any time that the inode on which
we are operating is a snapshot.
Help and review by: Jeff Roberson
Tested by: Peter Holm
MFC 233438:
Add a third flags argument to ffs_syncvnode to avoid a possible conflict
with MNT_WAIT flags that passed in its second argument.
Discussed with: kib
MFC 233629:
A refinement of change 232351 to avoid a race with a forcible unmount.
While we have a snapshot vnode unlocked to avoid a deadlock with another
inode in the same inode block being updated, the filesystem containing
it may be forcibly unmounted. When that happens the snapshot vnode is
revoked. We need to check for that condition and fail appropriately.
gibbs [Wed, 28 Mar 2012 19:40:58 +0000 (19:40 +0000)]
MFC Revision 233465
Correct failure to attach the PV block front device on Citrix
XenServer configurations that advertise the multi-page ring extension,
but only allow a single page of ring space.
sys/dev/xen/blkfront/blkfront.c:
If only one page of ring space is being used, do not publish
in the XenStore the number of pages in use (1), via either
of the supported multi-page ring extension schemes.
Single page operation is the same with or without the
ring-page extension being negotiated. Relying on the
legacy behavior avoids an incompatible difference in how
the two ring-page extension schemes that are out in the
wild, deal with the base case of a single page. The
Amazon/Red Hat drivers use the same XenStore variable as
if the extension was not negotiated. The Citrix drivers
assume the new ring reference XenStore variables will be
available
Reported by: Oliver Schonefeld <schonefeld@ids-mannheim.de>
zec [Wed, 28 Mar 2012 12:45:35 +0000 (12:45 +0000)]
MFC r232517:
Change SYSINIT priorities so that ip_mroute_modevent() is executed
before vnet_mroute_init(), since vnet_mroute_init() depends on mfchashsize
tunable to be set, and that is done in in ip_mroute_modevent().
Apparently I broke that ordering with r208744 almost 2 years ago...
mav [Wed, 28 Mar 2012 11:37:06 +0000 (11:37 +0000)]
MFC r232207, r232454:
Rework CPU load balancing in SCHED_ULE:
- In sched_pickcpu() be more careful taking previous CPU on SMT systems.
Do it only if all other logical CPUs of that physical one are idle to avoid
extra resource sharing.
- In sched_pickcpu() change general logic of CPU selection. First
look for idle CPU, sharing last level cache with previously used one,
skipping SMT CPU groups. If none found, search all CPUs for the least loaded
one, where the thread with its priority can run now. If none found, search
just for the least loaded CPU.
- Make cpu_search() compare lowest/highest CPU load when comparing CPU
groups with equal load. That allows to differentiate 1+1 and 2+0 loads.
- Make cpu_search() to prefer specified (previous) CPU or group if load
is equal. This improves cache affinity for more complicated topologies.
- Randomize CPU selection if above factors are equal. Previous code tend
to prefer CPUs with lower IDs, causing unneeded collisions.
- Rework periodic balancer in sched_balance_group(). With cpu_search()
more intelligent now, make balansing process flat, removing recursion
over the topology tree. That fixes double swap problem and makes load
distribution more even and predictable.
All together this gives 10-15% performance improvement in many tests on
CPUs with SMT, such as Core i7, for number of threads is less then number
of logical CPUs. In some tests it also gives positive effect to systems
without SMT.
mav [Wed, 28 Mar 2012 10:15:42 +0000 (10:15 +0000)]
MFC r232852:
Tune cpuset macros to optimize cases when CPU_SETSIZE fits into single
machine word. For example, it turns CPU_SET() into expected shift and OR,
removing extra shift, AND and additional index on memory access.
Generated code checked for kernel (optimized) and user-level (unoptimized)
cases with GCC and CLANG.
jkim [Tue, 27 Mar 2012 23:59:48 +0000 (23:59 +0000)]
MFC: r233042, r233054, r233056, r233187
- Do not unnecessarily clear display memory when switching modes.
- Remove unnecessary static variable initializations and duplicate codes.
- Consistently use bcopy(9) over memcpy(9).
- Save and restore linear frame buffer between suspend and resume.
yongari [Mon, 26 Mar 2012 05:14:04 +0000 (05:14 +0000)]
MFC r232951,232953,233158:
r232951:
fxp(4) does not handle deferred dma map loading. Tell
bus_dmamap_load(9) that it should return immediately with error
when there are insufficient mapping resources.
r232953:
Fix white space nits.
r233158:
Do not change current media when driver is already running. If
driver is running driver would have already completed flow control
configuration. This change removes unnecessary media changes in
controller reconfiguration cases such that it does not trigger link
reestablishment for configuration change requests like promiscuous
mode change.
Reported by: Many
Tested by: Mike Tancsa <mike <> sentex dot net>
yongari [Mon, 26 Mar 2012 04:47:06 +0000 (04:47 +0000)]
MFC r232849-232850:
r232849:
Show PCI bus speed and width as well as running mode of PCI-X
device in device attach. This would help to narrow down issue to a
specific controller and operating mode of the controller.
While I'm here rename BGE_MISCCFG_BOARD_ID with
BGE_MISCCFG_BOARD_ID_MASK.
r232850:
Make if_ierrors updated whenever any of the following counters are
updated.
o Number of times NIC ran out of RX buffer descriptors
o Number of inbound packet errors
o Number of inbound packets that were chosen to be discarded
Previously only the discarded packet counter was used to update
if_ierrors. This change fixes wrong if_ierrors counter on
BCM570[0-4] controllers. For BCM5705 and later controllers bge(4)
already correctly counted it.
yongari [Mon, 26 Mar 2012 04:36:22 +0000 (04:36 +0000)]
MFC r232848:
Add workaround for PCI-X BCM5704 controller that live behind
AMD-8131 PCI-X bridge. The bridge seems to reorder write access to
mailbox registers such that it caused watchdog timeouts by
out-of-order TX completions.
Tested by: Michael L. Squires <mikes <> siralan dot org >
yongari [Mon, 26 Mar 2012 04:27:01 +0000 (04:27 +0000)]
MFC r232246:
Prefer RL_GMEDIASTAT register to RGEPHY_MII_SSR register to
extract a link status of PHY when parent driver is re(4).
RGEPHY_MII_SSR register does not seem to report correct PHY status
on some integrated PHYs used with re(4).
Unfortunately, RealTek PHYs have no additional information to
differentiate integrated PHYs from external ones so relying on PHY
model number is not enough to know that. However, it seems
RGEPHY_MII_SSR register exists for external RealTek PHYs so
checking parent driver would be good indication to know which PHY
was used. In other words, for non-re(4) controllers, the PHY is
external one and its revision number is greater than or equal to 2.
This change fixes intermittent link UP/DOWN messages reported on
RTL8169 controller.
Also, mii_attach(9) is tried after setting interface name since
rgephy(4) have to know parent driver name.
yongari [Mon, 26 Mar 2012 03:54:19 +0000 (03:54 +0000)]
MFC r232145:
Use correct Config registers for RTL8139 family. Unlike RTL8168 and
RTL810x family , RTL8139 has different register map for Config
registers.
While here, follow the lead of re(4) in WOL configuration.
- Disable WOL_UCAST and WOL_MCAST capabilities by default.
- Config5 register write does not need to unlock EEPROM access
on RTL8139 family but unlocking EEPROM access does not affect
its operation and make it consistent with re(4).
Reported by: Matt Renzelmann mjr <> cs dot wisc dot edu
yongari [Mon, 26 Mar 2012 03:45:46 +0000 (03:45 +0000)]
MFC r232019,232021,232025,232027,232029,232031,232040:
r232019:
Give hardware chance to drain active DMA cycles.
r232021:
If there are not enough RX buffers, release partially allocated RX
buffers.
r232025:
Introduce sf_ifmedia_upd_locked() and have driver reset PHY before
switching to selected media. While here, set if_drv_flags before
switching to selected media.
r232027:
No need to reprogram hardware RX filter when driver is not running.
r232029:
Remove taskqueue based MII stat change handler.
Driver does not need deferred link state change processing.
While I'm here, do not report current link status if interface is
not UP.
r232031:
With r232015, sf(4) gets correct speed/duplex of established link.
Add more strict speed check in sf_miibus_statchg() and do not touch
MAC config registers when driver lost a link.
r232040:
Add check for IFF_DRV_RUNNING flag after serving an interrupt and
don't give RX path more priority than TX path.
Also remove infinite loop in interrupt handler and limit number of
iteration to 32. This change addresses system load fluctuations
under high network load.
yongari [Mon, 26 Mar 2012 02:40:54 +0000 (02:40 +0000)]
MFC r232015-232016:
r232015:
Add Seeq Technology 80220 PHY support to smcphy(4). This PHY is
found on Adaptec AIC-6915 Starfire ethernet controller.
While here, use status register to know resolved speed/duplex.
With this change, sf(4) correctly reports speed/duplex of
established link.
jilles [Sun, 25 Mar 2012 20:03:13 +0000 (20:03 +0000)]
MFC r233132: fts(3): Mention that FTS_NOCHDIR imposes {PATH_MAX} limits on
the returned pathnames.
With the current API (no *at functions), FTS_NOCHDIR requires that the
fts_accpath start with the original path passed to fts_open(); therefore,
the depth that can be reached is limited by the {PATH_MAX} constraint on
this pathname.
andreast [Sun, 25 Mar 2012 18:42:19 +0000 (18:42 +0000)]
MFC: r232482
Add support for PWM controlled fans. I found these fans on my PowerMac9,1.
These fans are not located under the same node as the the RPM controlled ones,
So I had to adapt the current source to parse and fill the properties correctly.
To control the fans we can set the PWM ratio via sysctl between 20 and 100%.
marius [Sun, 25 Mar 2012 16:24:42 +0000 (16:24 +0000)]
MFC: r233105
Declare some variables static in order to reduce the object size and
redo r232822 (MFC'ed to stable/9 in r232962) in a less hackish way.
The latter now no longer breaks compiling the x86 boot2 with clang.
edwin [Sun, 25 Mar 2012 02:18:22 +0000 (02:18 +0000)]
MFC of r233445, tzdata2012a
- Updates to various locations in Antarctica.
- Armenia will abolish DST this year.
- Not only Samoa has moved to UTC+14, also Fakaofo did.
- There will be a leap second in 30 June 2012.
- Historical updates of 1918 to Canada, Winn, Regina, Edm, Vanc, Creston.
- Chili stays on DST until 28 April 2012
- The Falkland islands will stay on DST this year.
kib [Sun, 25 Mar 2012 00:44:54 +0000 (00:44 +0000)]
MFC r232834:
In ffs_syncvnode(), pass boolean false as second argument of ffs_update().
Synchronous inode block update is not needed for MNT_LAZY callers (syncer),
and since waitfor values are not zero, code did unneccessary synchronous
update.
jhb [Fri, 23 Mar 2012 20:58:18 +0000 (20:58 +0000)]
MFC 232362,232363,232463,232553,232569,232571:
- Sort function prototypes.
- Update the documentation on pci_get/set_powerstate(). These methods are
not ACPI-specific at all, but deal with PCI power states. Also,
pci_set_powerstate() fails with EOPNOTSUPP if a request is made that the
underlying device does not support rather than falling back to somehow
setting D0.
- Document the following routines: pci_alloc_msi(), pci_alloc_msix(),
pci_find_cap(), pci_get_max_read_req(), pci_get_vpd_ident(),
pci_get_vpd_readonly(), pci_msi_count(), pci_msix_count(),
pci_pending_msix(), pci_release_msi(), pci_remap_msix(), and
pci_set_max_read_req().
- Group the functions into five sub-sections: raw configuration access,
locating devices, device information, device configuration, and
message signaled interrupts.
- Discourage use of pci_disable_io() and pci_enable_io() in device drivers.
The PCI bus driver handles this automatically as resources are activated.
jhb [Fri, 23 Mar 2012 20:47:25 +0000 (20:47 +0000)]
MFC 232403,232667:
- Add a bus_dma tag to each PCI bus that is a child of a Host-PCI bridge.
The tag enforces a single restriction that all DMA transactions must not
cross a 4GB boundary. Note that while this restriction technically only
applies to PCI-express, this change applies it to all PCI devices as it
is simpler to implement that way and errs on the side of caution.
- Add a softc structure for PCI bus devices to hold the bus_dma tag and
a new pci_attach_common() routine that performs actions common to the
attach phase of all PCI bus drivers. Right now this only consists of
a bootverbose printf and the allocate of a bus_dma tag if necessary.
- Adjust all PCI bus drivers to allocate a PCI bus softc and to call
pci_attach_common() from their attach routines.
jhb [Fri, 23 Mar 2012 18:29:09 +0000 (18:29 +0000)]
MFC 232401:
Similar to the fixes in 226967 and 226987, purge any name cache entries
associated with the previous vnode (if any) associated with the target of
a rename(). Otherwise, a lookup of the target pathname concurrent with a
rename() could re-add a name cache entry after the namei(RENAME) lookup
in kern_renameat() had purged the target pathname.
jhb [Fri, 23 Mar 2012 18:03:04 +0000 (18:03 +0000)]
MFC 232360:
Add pci_save_state() and pci_restore_state() wrappers around
pci_cfg_save() and pci_cfg_restore() for device drivers to use when
saving and restoring state (e.g. to handle device-specific resets).
jhb [Fri, 23 Mar 2012 17:22:43 +0000 (17:22 +0000)]
MFC 232570,232754:
Fix boot2 to handle boot config files that only contain a custom path to
a loader or kernel. Specifically, kname cannot be pointed at cmd[] since
it's value is change to be an empty string after the initial call to
parse, and cmd[]'s value can be changed (thus losing a prior setting for
kname) due to user input at the boot prompt. While here, ensure that that
initial boot config file text is nul-terminated, that ops is initialized
to zero, and that kname is always initialized to a valid string.
In addition, include other changes to ensure boot2 still builds with
Clang.
jimharris [Fri, 23 Mar 2012 16:59:03 +0000 (16:59 +0000)]
MFC r233371:
Call xpt_bus_register during attach context, then freeze and do not release
until domain discovery is complete. This fixes an isci(4) bug on FreeBSD 7.x
where devices weren't always appearing after boot without an explicit rescan.
Sponsored by: Intel
Reported and tested by: <rpokala at panasas dot com>
Reviewed by: scottl
Approved by: scottl
mav [Fri, 23 Mar 2012 16:17:46 +0000 (16:17 +0000)]
MFC r232793:
Revert r175376 and tune cpufreq(4) frequency comparison logic instead.
Instead of using 25MHz equality threshold, look for the nearest value when
handling dev.cpu.0.freq sysctl and for exact match when it is expected.
ACPI may report extra level with frequency 1MHz above the nominal to
control Intel Turbo Boost operation. It is not a bug, but feature:
dev.cpu.0.freq_levels: 2934/106000 2933/95000 2800/82000 ...
In this case value 2933 means 2.93GHz, but 2934 means 3.2-3.6GHz.
I've found that my Core i7 based systems have Intel Turbo Boost not used
by default and without this change it was absolutely invisible and hard
to control.
mav [Fri, 23 Mar 2012 16:12:52 +0000 (16:12 +0000)]
MFC r232717:
Be more polite when setting state->nextevent inside cpu_new_callout().
Hardclock is not the only who wakes idle CPU since kdtrace cyclic addition.
jhb [Fri, 23 Mar 2012 12:05:51 +0000 (12:05 +0000)]
MFC 232218:
Clear the a device's description string anytime it's driver changes.
Descriptions are specific to drivers and we don't change drivers on attached
devices. This fixes a few places where we were not clearing the description
when detaching a driver (e.g. if device_attach() failed). While here, fix
a few other nits:
- Remove spurious call to remove a device's driver from
devclass_driver_deleted(). device_detach() removes it already.
- Fix a typo.
kib [Fri, 23 Mar 2012 11:26:54 +0000 (11:26 +0000)]
MFC r231949:
Fix found places where uio_resid is truncated to int.
Add the sysctl debug.iosize_max_clamp, enabled by default. Setting the
sysctl to zero allows to perform the SSIZE_MAX-sized i/o requests from
the usermode.
MFC r232493:
Remove unneeded cast to u_int. The values as small enough to fit into
int, beside the use of MIN macro which performs type promotions.
MFC r232494:
Instead of incomplete handling of read(2)/write(2) return values that
does not fit into registers, declare that we do not support this case
using CTASSERT(), and remove endianess-unsafe code to split return value
into td_retval.
While there, change the style of the sysctl debug.iosize_max_clamp
definition.
MFC r232495:
pipe_read(): change the type of size to int, and remove signed clamp.
pipe_write(): change the type of desiredsize back to int, its value fits.
dim [Fri, 23 Mar 2012 11:01:04 +0000 (11:01 +0000)]
MFC r232749:
Partially undo r228693, by removing NO_WFORMAT.clang in fsdb's Makefile,
and fixing the format string in sbin/fsdb/fsdbutil.c instead.
Note the remark "Work around a problem with format string warnings and
ntohs macros" was actually incorrect. The DIP(dp, di_nlink) macro
invocation actually returned an int, due to its ternary expression, even
though the di_nlink members of struct ufs1_dinode and struct ufs2_dinode
are both defined as int16_t.
dim [Fri, 23 Mar 2012 10:55:19 +0000 (10:55 +0000)]
MFC r232745:
Add casts to __uint16_t to the __bswap16() macros on all arches which
didn't already have them. This is because the ternary expression will
return int, due to the Usual Arithmetic Conversions. Such casts are not
needed for the 32 and 64 bit variants.
While here, add additional parentheses around the x86 variant, to
protect against unintended consequences.
jhb [Thu, 22 Mar 2012 21:07:54 +0000 (21:07 +0000)]
MFC 230547:
Add a timeout on positive name cache entries in the NFS client. That is,
we will only trust a positive name cache entry for a specified amount of
time before falling back to a LOOKUP RPC, even if the ctime for the file
handle matches the cached copy in the name cache entry. The timeout is
configured via a new 'nametimeo' mount option and defaults to 60 seconds.
It may be set to zero to disable positive name caching entirely.
jhb [Thu, 22 Mar 2012 20:51:35 +0000 (20:51 +0000)]
MFC 232116:
Adjust the nfs_skip_wcc_data_onerr setting so that it does not block
post-op attributes for ENOENT errors now that the name caching logic
depends on working post-op attributes.
pfg [Thu, 22 Mar 2012 20:31:52 +0000 (20:31 +0000)]
MFC: 232703
Add support for ns timestamps and birthtime to the ext2/3 driver.
When using big inodes there is sufficient space in ext3 to
keep extra resolution and birthtime (creation) timestamps.
We now support the extra timestamps silently when the inode
is big enough.
dim [Thu, 22 Mar 2012 09:50:16 +0000 (09:50 +0000)]
MFC r233195:
Fix the following warning from clang trunk:
usr.sbin/dconschat/dconschat.c:163:65: error: format specifies type 'size_t' (aka 'unsigned int') but the argument has type 'off_t' (aka 'long long') [-Werror,-Wformat]
snprintf(buf, PAGE_SIZE, "\r\n[dconschat reset target(addr=0x%zx)...]\r\n", dc->reset);
~~^ ~~~~~~~~~
%llx
Silence this by casting dc->reset to intmax_t, and using the appropriate
length modifier. While here, wrap the line to a 80 character margin.
dim [Thu, 22 Mar 2012 09:47:10 +0000 (09:47 +0000)]
MFC r233193:
Add explicit braces to avoid dangling else in stl_tree.h. This silences
the following warning produced by clang trunk:
In file included from /usr/src/sbin/devd/devd.cc:91:
In file included from /usr/obj/usr/src/tmp/usr/include/c++/4.2/map:64:
/usr/obj/usr/src/tmp/usr/include/c++/4.2/bits/stl_tree.h:987:2: error: add explicit braces to avoid dangling else [-Werror,-Wdangling-else]
else
^
jhb [Wed, 21 Mar 2012 20:50:15 +0000 (20:50 +0000)]
MFC 230394,230441,230489,230552,232420:
Close a race in NFS lookup processing that could result in stale name cache
entries on one client when a directory was renamed on another client. The
root cause for the stale entry being trusted is that each per-vnode nfsnode
structure has a single 'n_ctime' timestamp used to validate positive name
cache entries. However, if there are multiple entries for a single vnode,
they all share a single timestamp. To fix this, extend the name cache
to allow filesystems to optionally store a timestamp value in each name
cache entry. The NFS clients now fetch the timestamp associated with
each name cache entry and use that to validate cache hits instead of the
timestamps previously stored in the nfsnode. Another part of the fix is
that the NFS clients now use timestamps from the post-op attributes of
RPCs when adding name cache entries rather than pulling the timestamps out
of the file's attribute cache. The latter is subject to races with other
lookups updating the attribute cache concurrently.
glebius [Wed, 21 Mar 2012 09:19:23 +0000 (09:19 +0000)]
Merge 232685 from head:
Merge from OpenBSD:
revision 1.146
date: 2010/05/12 08:11:11; author: claudio; state: Exp; lines: +2 -3
bzero() the full compressed update struct before setting the values.
This is needed because pf_state_peer_hton() skips some fields in certain
situations which could result in garbage beeing sent to the other peer.
This seems to fix the pfsync storms seen by stephan@ and so dlg owes me
a whiskey.
I didn't see any storms, but this definitely fixes a useless memory
allocation on the receiving side, due to non zero scrub_flags field
in a pfsync_state_peer structure.
glebius [Wed, 21 Mar 2012 07:22:27 +0000 (07:22 +0000)]
Merge from head:
r227785:
- Reduce severity for all ARP events, that can be triggered from remote
machine to LOG_NOTICE. Exception left to "using my IP address".
- Fix multicast ARP warning: add newline and also log the bad MAC address.
Tested by: Alexander Wittig <wittigal msu.edu>
r227790:
Be more informative for "unknown hardware address format" message.
tuexen [Tue, 20 Mar 2012 19:22:44 +0000 (19:22 +0000)]
MFC r232866:
This fixes PR 165210. Basically we just
add in the netgraph interface to the list of
acceptable interfaces. A todo at the next
IETF code blitz, though is we need to review
why we screen interfaces, there was a reason ;-).
From rrs@.
jpaetzel [Mon, 19 Mar 2012 23:17:26 +0000 (23:17 +0000)]
MFC: 232898
Improve ZFS exporting functionality, only export pools which are on a
specific device we happen to be writing to. This fixes an issue when
running pc-sysinstall on a running system which needs ZFS and the main
disk gets exported.
jpaetzel [Mon, 19 Mar 2012 22:52:19 +0000 (22:52 +0000)]
MFC: 232681
Work around broken BIOS memory reporting
Andrzej has a machine with 32GB of RAM, but only 16GB is reported by the
smbios.memory.enabled. Thus, use the greater of hw.realmem and the smbios
value.
Reported by: Andrzej Tobola <ato of iem pw edu pl>
jpaetzel [Mon, 19 Mar 2012 22:41:53 +0000 (22:41 +0000)]
MFC 227470:
Report the amount of memory from smbios data if provided.
This should get the correct memory size even if a 32-bit image is running
on a machine with > 4GB of memory. This can be useful is using a 32-bit
installer on a machine which will eventually run a 64-bit image.
dim [Mon, 19 Mar 2012 08:10:23 +0000 (08:10 +0000)]
MFC r232894:
Pull in r145194 from upstream clang trunk:
Make our handling of MMX x SSE closer to what gcc does:
* Enabling sse enables mmx.
* Disabling (-mno-mmx) mmx, doesn't disable sse (we got this right already).
* The order in not important. -msse -mno-mmx is the same as -mno-mmx -msse.
Some configure scripts depend on this.
PR: i386/165968
MFC r232933:
Update comments and CFLAGS in sys/conf/kern.mk, introduced in r221879,
to match reality: clang does _not_ disable SSE automatically when
-mno-mmx is used, you have to specify -mno-sse explicitly.
Note this was the case even before r232894, which only makes a change in
the 'positive' flag case; e.g. when you specify -msse, MMX gets enabled
too.
mav [Mon, 19 Mar 2012 07:24:26 +0000 (07:24 +0000)]
MFC r230845:
Make sound(4) more flexible in setting soft buffer and block sizes when
hardware imposes strict limitations on hard buffer and block sizes.
Previous code set soft buffer to be no smaller then hard buffer. On some
cards with fixed 64K physical buffer that caused up to 800ms play latency.
New code allows to set soft buffer size down to just two blocks of the hard
buffer and to not write more then that size ahead to the hardware buffer.
As result of that change I was able to reduce full practically measured
record-playback loop delay in those conditions down to only about 115ms
with theoretical playback latency of only about 50ms.
New code works fine for both vchans and direct cases. In both cases sound(4)
tries to follow hw.snd.latency_profile and hw.snd.latency values and
application-requested buffer and block sizes as much as limitation of two
hardware blocks allows.
projects / FreeBSD / stable / 9.git / log