Warner Losh [Wed, 26 Aug 2015 17:10:43 +0000 (17:10 +0000)]
After consultations with the arm community, don't define softfp for
armv6. It's too ambiguous. We do use the softfp ABI for the moment on
armv6, but we allow floating point register use (and the compilers
will generate it). This is too ambiguous to use it as a decider for
which algorithms to use on the platform. Err on the side of caution
and not define it.
Zbigniew Bodek [Wed, 26 Aug 2015 12:32:46 +0000 (12:32 +0000)]
Fix race condition in its_cmd_send()
its_cmd_send() can be called by multiple CPUs simultaneously.
After the command is pushed to ITS command ring the completion
status is polled using global pointer to the next free ring slot.
Use copied pointer and provide correct locking to avoid spurious
pointer value when concurrent access occurs.
Obtained from: Semihalf
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3436
Andrew Turner [Wed, 26 Aug 2015 11:36:23 +0000 (11:36 +0000)]
Add an option to select which SoCs we are building for. It is intended to
be used with any SoC specific drivers, for example a ThunderX nic driver
would use something like the following in files.arm64:
Bryan Drewery [Wed, 26 Aug 2015 03:44:48 +0000 (03:44 +0000)]
Fix filemon locking races.
Convert filemon_lock and struct filemon* lock to sx(9), rather than a
self-rolled reader-writer lock, and hold it for the entire time needed.
At least filemon_lock_write() was not checking for active readers when
it would successfully return with the write lock "held". This led to
a race with reading entries from filemon_inuse as they were removed. This
could be seen with QUEUE_MACRO_DEBUG enabled, causing -1 to be read as an
entry rather than a valid struct filemon*.
Fixing filemon_lock_write() to check readers was insufficient to fix the
races.
sx(9) was used as the lock could be held while taking proctree_lock and sleeping
in fo_write.
Warner Losh [Tue, 25 Aug 2015 17:11:49 +0000 (17:11 +0000)]
Add softfp to MACHINE_CPU more often when we're compiling for soft
float targets. It is added for booke on powerpc and all arm with hf in
the string. Also add arm to all arm builds and armv6 to armv6 and
newer builds.
Andrew Turner [Tue, 25 Aug 2015 17:02:28 +0000 (17:02 +0000)]
Use kdb_jmpbuf and setjmp to handle accessing invalid addresses in the
debugger memory access functions. This allows us to correctly reenter into
the debugger on failure.
Alan Cox [Tue, 25 Aug 2015 01:01:25 +0000 (01:01 +0000)]
Testing whether a page is dirty does not require the page lock. Moreover,
it may involve a pmap operation that iterates over the page's PV list, so
unnecessarily holding the page lock is undesirable.
Conrad Meyer [Mon, 24 Aug 2015 19:32:03 +0000 (19:32 +0000)]
Import ioat(4) driver
I/OAT is also referred to as Crystal Beach DMA and is a Platform Storage
Extension (PSE) on some Intel server platforms.
This driver currently supports DMA descriptors only and is part of a
larger effort to upstream an interconnect between multiple systems using
the Non-Transparent Bridge (NTB) PSE.
For now, this driver is only built on AMD64 platforms. It may be ported
to work on i386 later, if that is desired. The hardware is exclusive to
x86.
Further documentation on ioat(4), including API documentation and usage,
can be found in the new manual page.
Bring in a test tool, ioatcontrol(8), in tools/tools/ioat. The test
tool is not hooked up to the build and is not intended for end users.
Bjoern A. Zeeb [Mon, 24 Aug 2015 18:57:32 +0000 (18:57 +0000)]
When forking a child process with PMC_F_DESCENDANTS set in pmc_attach()
in the parent, we will inherit the pmcids but cannot execute any operations
on them in the child. The reason for this is that pmc_find_pmc() only
tries to find the current process on the owners hash list, but given the
child does not own the attachment, we cannot find it.
Thus, in case the initial lookup fails, try to find the pmc_process state
affiliated with the child process, lookup the pmc from there using the
row index, and get the owner process from that pmc.
Then continue as normal and lookup the pmc context of the owner (process).
This allows us to call, e.g., pmc_start() in the child process before we
start the work there, but to collect the accumulated results later in
the parent.
Make vfs_unmountall() unmount /dev after /, not before. The only
reason this didn't result in an unclean shutdown is that devfs ignores
MNT_FORCE flag.
Reviewed by: kib@
MFC after: 1 month
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3467
Andrew Turner [Mon, 24 Aug 2015 12:17:15 +0000 (12:17 +0000)]
Fix libproc on architectures that don't need the program counter to be
adjusted. This seems to be the case on all non-x86 architectures libproc
supports.
Reviewed by: kib
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3465
Andriy Gapon [Mon, 24 Aug 2015 09:48:50 +0000 (09:48 +0000)]
MFV (partial) r286889: 5692 expose the number of hole blocks in a file
FreeBSD porting notes:
- only kernel-side changes are merged
- the new ioctl is not actually implemented yet
- thus, the goal is to synchronize DMU code
https://www.illumos.org/issues/5692
we would like to expose the number of hole (sparse) blocks in a file.
this can be useful to for example if you want to fill in the holes with
some data; knowing the number of holes in advances allows you to report
progress on hole filling. We could use SEEK_HOLE to do that but it would
be O(n) where n is the number of holes present in the file.
Author: Max Grossman <max.grossman@delphix.com>
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Boris Protopopov <bprotopopov@hotmail.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Julien Charbon [Mon, 24 Aug 2015 09:30:27 +0000 (09:30 +0000)]
Revert r284245: "Fix a callout race condition introduced in TCP
timers callouts with r281599."
r281599 fixed a TCP timer race condition, but due a callout(9) bug
it also introduced another race condition workaround-ed with r284245.
The callout(9) bug being fixed with r286880, we can now revert the
workaround (r284245).
Andriy Gapon [Mon, 24 Aug 2015 08:10:52 +0000 (08:10 +0000)]
account for ashift when gathering buffers to be written to l2arc device
The change that introduced the L2ARC compression support also introduced
a bug where the on-disk size of the selected buffers could end up larger
than the target size if the ashift is greater than 9. This was because
the buffer selection could did not take into account the fact that
on-disk size could be larger than the in-memory buffer size due to
the alignment requirements.
At the moment b_asize is a misnomer as it does not always represent the
allocated size: if a buffer is compressed, then the compressed size is
properly rounded (on FreeBSD), but if the compression fails or it is not
applied, then the original size is kept and it could be smaller than what
ashift requires.
For the same reasons arcstat_l2_asize and the reported used space
on the cache device could be smaller than the actual allocated size
if ashift > 9. That problem is not fixed by this change.
This change only ensures that l2ad_hand is not advanced by more
than target_sz. Otherwise we would overwrite active (unevicted)
L2ARC buffers. That problem is manifested as growing l2_cksum_bad
and l2_io_error counters.
This change also changes 'p' prefix to 'a' prefix in a few places
where variables represent allocated rather than physical size.
The resolved problem could also result in the reported allocated size
being greater than the cache device's capacity, because of the
overwritten buffers (more than one buffer claiming the same disk
space).
Ed Schouten [Mon, 24 Aug 2015 07:49:27 +0000 (07:49 +0000)]
Sync HPA and VPA implementations with CUP.
After fixing the 16-bits integer arithmetic overflow in 286981, we
should also make sure to fix the VPA sequence. Bring HPA and VPA in sync
with how we now implement CUP.
Hiroki Sato [Mon, 24 Aug 2015 05:21:49 +0000 (05:21 +0000)]
- Deprecate IN6_IFF_NODAD. It was used to prevent DAD on a loopback
interface but in6if_do_dad() already had a check for IFF_LOOPBACK.
- Remove in6if_do_dad() check in in6_broadcast_ifa(). An address
which needs DAD always has IN6_IFF_TENTATIVE there.
- in6if_do_dad() now returns EAGAIN when the interface is not ready
since DAD callout handler ignores such an interface.
- In DAD callout handler, mark an address as IN6_IFF_TENTATIVE
when the interface has ND6_IFF_IFDISABLED. And Do IFF_UP and
IFF_DRV_RUNNING check consistently when DAD is required.
- draft-ietf-6man-enhanced-dad is now published as RFC 7527.
Warner Losh [Mon, 24 Aug 2015 00:03:51 +0000 (00:03 +0000)]
We need to add the soft float to the CFLAGS always, not just when
NO_CPUCFLAGS is not defined since it is part of the ABI as we've
defined it, not just a nice optimization.
Jilles Tjoelker [Sun, 23 Aug 2015 20:44:53 +0000 (20:44 +0000)]
sh: Don't create bad parse result when postponing a bad substitution error.
An invalid substitution like ${var@} does not cause a parse error but is
stored in the intermediate representation, to be written as part of the
error message. If there is a CTL* byte in the stored part, this confuses
some code such as the code to skip an unused alternative such as in
${var-alternative}.
Update svnlite(1) commit template. This makes it possible to set the
"freebsd-sponsored-by" and "freebsd-mfc-after" in ~/.subversion/config,
in the "[miscellany]" section - just like you already can with
devel/subversion port/package.
MFC after: 1 month
Sponsored by: The FreeBSD Foundation
Xin LI [Sun, 23 Aug 2015 08:22:00 +0000 (08:22 +0000)]
Although the website [1] says US-ASCII is accepted, ISO-8859-1 seems to
be the only accepted character that is close enough to US-ASCII (and
this is also done on OpenBSD; the alternative would be to use UTF-8).
Adrian Chadd [Sun, 23 Aug 2015 01:17:52 +0000 (01:17 +0000)]
Reset the channel to the first available channel if the interface
is configured on a channel that isn't valid in the new operating mode.
This isn't strictly true - it should find the first channel that is
available for the given operating mode.
However, I think defaulting to the first channel is fine - it's typically
available for all modes.
If someone would like to correctly implement this feature - try to
find a channel that is valid for the given operating mode and error
out if we can't find one.
This prevents various NICs (eg wpi(4)) from throwing a firmware error.
Mark Murray [Sat, 22 Aug 2015 12:59:05 +0000 (12:59 +0000)]
Make the UMA harvesting go away completely if not wanted. Default to "not wanted".
Provide and document the RANDOM_ENABLE_UMA option.
Change RANDOM_FAST to RANDOM_UMA to clarify the harvesting.
Remove RANDOM_DEBUG option, replace with SDT probes. These will be of
use to folks measuring the harvesting effect when deciding whether to
use RANDOM_ENABLE_UMA.
Requested by: scottl and others.
Approved by: so (/dev/random blanket)
Differential Revision: https://reviews.freebsd.org/D3197
Justin Hibbits [Sat, 22 Aug 2015 07:20:03 +0000 (07:20 +0000)]
Enhance book-e pmap for 36-bit physaddr
Summary:
This is (probably step 1) of enhancing the book-e pmap to support the full
36-bit physical address space on Freescale e500 and e5500 cores.
Thus far it has only been regression tested on one platform. Since I only have
one other Book-E platform (e5500), that needs work beyond this, I haven't yet
tested it on this.
Test Plan: Regression tested on my RouterBoard RB800.
Justin Hibbits [Sat, 22 Aug 2015 05:50:18 +0000 (05:50 +0000)]
Create a RouterBoard platform and use it to create a flash map
Summary:
The RouterBoard uses a predefined partition map which doesn't exist in the fdt.
This change allows overriding the fdt slicer with a custom slicer, and uses this
custom slicer to define the flash map on the RouterBoard RB800.
D3305 converts the mpc85xx platform into a base class, so that systems based on
the mpc85xx platform can add their own overrides. This change builds on D3305,
and creates a RouterBoard (RB800) platform to initialize the slicer override.
Justin Hibbits [Sat, 22 Aug 2015 03:29:12 +0000 (03:29 +0000)]
Make the mpc85xx platform a kobj base class.
Summary:
Some systems are based around mpc85xx, but need special initialization. By
making the mpc85xx platform a base class, these systems can be platform
subclasses, and perform board-specific initialization in addition to the mpc85xx
initialization.
Test Plan:
Tested on my RB800. A platform class was created, and will be committed
separately.
Roger Pau Monné [Fri, 21 Aug 2015 15:57:57 +0000 (15:57 +0000)]
preload_search_info: make sure mod is set
Add a check to preload_search_info to make sure mod is set. Most of the
callers of preload_search_info don't check that the mod parameter is
set, which can cause page faults. While at it, remove some now unnecessary
checks before calling preload_search_info.
Roger Pau Monné [Fri, 21 Aug 2015 15:53:08 +0000 (15:53 +0000)]
xen: allow disabling PV disks and nics
Introduce two new loader tunnables that can be used to disable PV disks and
PV nics at boot time. They default to 0 and should be set to 1 (or any
number different than 0) in order to disable the PV devices: