head/sys/amd64/amd64/support.S: unroll loop
unroll the loop in ENTRY(pagezero)
acc' to the submitter this results in a reproducible 1% perf
improvement under buildworld like workload
I validated correctness and run-testing, but not performance impact
adrian [Sun, 5 Apr 2015 02:57:02 +0000 (02:57 +0000)]
Add support for the MIPS74K SoC family performance counters events.
These are similar to the mips24k performance counters - some are
available on perfcnt0/3, some are available on perfcnt1/4.
However, the events aren't all the same.
* Add the events, named the same as from Linux oprofile.
* Verify they're the same as "MIPS32(R) 74KTM Processor Core Family
Software User's Manual"; Document Number: MD00519; Revision 01.05.
* Rename INSTRUCTIONS to something else, so it doesn't clash with
the alias INSTRUCTIONS. I'll try to tidy this up later; there
are a few other aliases to add and shuffle around.
Tested:
* QCA9558 SoC (AP135 board) - MIPS74Kc core (no FPU.)
* make universe; where it didn't fail for other reasons.
TODO:
* It'd be nice to support the four performance counters
in at least this hardware, rather than just two.
Summary:
Book-E and AIM trap.c are almost identical, except for a few bits. This is step
1 in unifying them.
This also renumbers EXC_DEBUG, to not conflict with AIM vector numbers. Since
this is the only one thus far that is used in the switch statement in trap(),
it's the only one renumbered. If others get added to the switch, which conflict
with AIM numbers, they should also be renumbered.
When checking the length of the mutual secret password the variable for
the secret password was used by mistake. This resulted in ctld never
warning about the length of the mutual secret being wrong even if it was.
fts: Don't return FTS_SLNONE if it's not a symlink (if race).
When following symlinks, fts returned FTS_SLNONE when fstatat(flag=0)
failed, but a subsequent fstatat(flag=AT_SYMLINK_NOFOLLOW) succeeded. This
incorrectly triggered if a filename existed to be read from the directory,
was deleted before the fstatat(flag=0) and created again after the
fstatat(flag=0).
Fix this by only returning FTS_SLNONE if the result from
fstatat(flag=AT_SYMLINK_NOFOLLOW) is actually a symlink. If it is not a
symlink, treat it as if fstatat(flag=0) succeeded.
Replace vm_fault()'s heuristic for automatic cache behind with a heuristic
that performs the equivalent of an automatic madvise(..., MADV_DONTNEED).
The current heuristic, even with the improvements that I made a few years
ago, is a good example of making the wrong trade-off, or optimizing for
the infrequent case. The infrequent case being reading a single file that
is much larger than memory using mmap(2). And, in this case, the page
daemon isn't the bottleneck; it's the I/O.
In all other cases, the current heuristic has too many false positives,
i.e., it caches too many pages that are later reused. To give one
example, thousands of pages are cached by the current heuristic during a
buildworld and all of them are reactivated before the buildworld
completes. In particular, clang reads source files using mmap(2) and
there are some relatively large source files in our source tree, e.g.,
sqlite, that are read multiple times. With the new heuristic, I see fewer
false positives and they have a much lower cost.
I actually tried something like this more than two years ago and it
didn't perform as well as the cache behind heuristic. However, that was
before the changes to the page daemon in late summer of 2013 and the
existence of pmap_advise(). In particular, with the page daemon doing
its work more frequently and in smaller batches, it now completes its
work while the application accessing the file is blocked on I/O.
Whereas previously, the page daemon appeared to hog the CPU for so long
that it caused "hiccups" in the application's execution.
Finally, I'll add that the elimination of cache pages is a prerequisite
for NUMA support.
andrew [Sat, 4 Apr 2015 09:07:31 +0000 (09:07 +0000)]
Add support for arm64 to the existing arm generic timer driver:
- Add macros to handle the differences in accessing these registers on arm
and arm64.
- Use the fdt data to detect if we are on an ARMv7 or ARMv8.
- Use the virtual timer by default on arm64, we may not have access to
the physical timer.
Differential Revision: https://reviews.freebsd.org/D2208
Reviewed by: emaste
Sponsored by: The FreeBSD Foundation
Previously, the driver was trying to blink the LED in the newstate
function, but that only gets called once (unlike OpenBSD's net80211
stack). Move the LED blinking to set_channel().
While there, don't try to set the channel when we switch to the SCAN
state. This is already accomplished by the set_channel() function.
boot1 EFI: reset the screen and select the best mode.
It's necessary to reset the screen to make sure any vendor pixels are
gone when we start boot1. In the Lenovo X1 (3rd gen), this is the
only way to clear the screen. Previously, the Lenovo logo would only
disappear after the kernel started scrolling the display.
After resetting the screen, EFI could put us in the worst LCD mode
(oversized characters), so we now find the largest mode we can use and
hope it's the most appropriate one (it's not trivial to tell what's
the correct LCD resolution at this point). It's worth noting that the
final stage loader has a 'mode' command that can be used to switch
text modes.
While there, enable the software cursor, just like in the legacy boot
mode.
andrew [Fri, 3 Apr 2015 19:33:26 +0000 (19:33 +0000)]
More ARM EABI object attributes in binutils. This adds support to binutils
to include the Unaligned Access and Floating-point Half-precision
attributes. the former marks ELF objects that may access ARMv6 style
unaligned data, the latter that the binary uses the VFPv3/Advanced SIMD
half-precision extension.
These may be emmitted by clang so it's best to print a warning when the
linker hits one of them.
dim [Fri, 3 Apr 2015 18:38:37 +0000 (18:38 +0000)]
Pull in r227115 from upstream clang trunk (by Ben Langmuir):
Fix assert instantiating string init of static variable
... when the variable's type is a typedef of a ConstantArrayType. Just
look through the typedef (and any other sugar). We only use the
constant array type here to get the element count.
This fixes an assertion failure when building the games/redeclipse port.
Make ZFS ARC track both KVA usage and fragmentation.
Even on Illumos, with its much larger KVA, ZFS ARC steps back if KVA usage
reaches certain threshold (3/4 on i386 or 16/17 otherwise). FreeBSD has
even less KVA, but had no such limit on archs with direct map as amd64.
As result, on machines with a lot of RAM, during load with very small user-
space memory pressure, such as `zfs send`, it was possible to reach state,
when there is enough both physical RAM and KVA (I've seen up to 25-30%),
but no continuous KVA range to allocate even single 128KB I/O request.
Address this situation from two sides:
- restore KVA usage limitations in a way the most close to Illumos;
- introduce new requirement for KVA fragmentation, specifying that we
should have at least one sequential KVA range of zfs_max_recordsize bytes.
Experiments show that first limitation done alone is not sufficient. On
machine with 64GB of RAM it is sometimes needed to drop up to half of ARC
size to get at leats one 1MB KVA chunk. Statically limiting ARC to half
of KVA/RAM is too strict, so second limitation makes it to work in cycles:
accumulate trash up to certain critical mass, do massive spring-cleaning,
and then start littering again. :)
andrew [Fri, 3 Apr 2015 09:35:52 +0000 (09:35 +0000)]
Add support for thread local storage on arm64 to the runtime linker. The
ABI specifies that, for R_AARCH64_TLSDESC relocations, we use the symbol
value, addend, and object tls offset to calculate the offset from the tls
base. We then cache this value for future reference.
Differential Revision: https://reviews.freebsd.org/D2183
Reviewed by: kib
Sponsored by: The FreeBSD Foundation
peter [Thu, 2 Apr 2015 23:12:18 +0000 (23:12 +0000)]
Remove redundant mtx_lock/unlock in ciss_name_device. This is a guaranteed
insta-panic on device add/remove. This is only called from the notify
thread which already holds the lock while calling this function.
When an mbuf allocation fails in the receive path, the mbuf containing the received packet is not sent to the host net
work stack and is reused again on the receive ring. Remaining received packets in the ring are not processed in that
invocation of bxe_rxeof() and defered to the task thread.
Speed up symbol lookup for the amd64 kernel modules.
Amd64 uses relocatable object files as the modules format. It is good
WRT not having unneeded overhead for PIC code, in particular, due to
absence of useless GOT and PLT. But the cost is that the module
linking process cannot use hash to speed up the symbol lookup, and
that each reference to the symbol requiring a relocation, instead of
single-place relocation in GOT.
Cache the successfull symbol lookup results in the module symbol
table, using the newly allocated SHN_FBSD_CACHED value from
SHN_LOOS-HIOS range as an indicator. The SHN_FBSD_CACHED together
with the non-existent definition of the found symbol are reverted
after successfull relocations, which is done under kld_sx lock, so it
should not be visible to other consumers of the symbol table.
Move i386/efi files to new home in efi/loader/arch/i386
This was not (and still is not) connected to the build, but the EFI
loader is in the process of being built for other than amd64 so these
files ought to live in their eventual MD location.
- dtucker@cvs.openbsd.org 2007/12/28 22:34:47
[clientloop.c]
Use the correct packet maximum sizes for remote port and agent forwarding.
Prevents the server from killing the connection if too much data is queued
and an excessively large packet gets sent. bz #1360, ok djm@.
The change was lost due to the the way the original upstream HPN patch
modified this code. It was re-adding the original OpenSSH code and never
was properly fixed to use the new value.
Extend fixes made in r278103 and r38754 by copying the complete packet
header and not only partial flags and fields. Firewalls can attach
classification tags to the outgoing mbufs which should be copied to
all the new fragments. Else only the first fragment will be let
through by the firewall. This can easily be tested by sending a large
ping packet through a firewall. It was also discovered that VLAN
related flags and fields should be copied for packets traversing
through VLANs. This is all handled by "m_dup_pkthdr()".
Regarding the MAC policy check in ip_fragment(), the tag provided by
the originating mbuf is copied instead of using the default one
provided by m_gethdr().
- Make interrupt resource optional: some upstream FDT blobs (e.g. TI's) do
not have interupt property in pl310 node. Interrupt is used only to
detect cache activity when L2 cache is disabled, it's not vital for
normal operations.
- Fix intrhook allocation/initialization
There are cases when gpioled nodes in DTS come from different sources
(e.g. standard Beaglebone Black LEDs in main DTS + shield LEDs in
overlay DTS) so instead of handling only first compatible node go
through all child nodes
o Use new function ip_fillid() in all places throughout the kernel,
where we want to create a new IP datagram.
o Add support for RFC6864, which allows to set IP ID for atomic IP
datagrams to any value, to improve performance. The behaviour is
controlled by net.inet.ip.rfc6864 sysctl knob, which is enabled by
default.
o In case if we generate IP ID, use counter(9) to improve performance.
o Gather all code related to IP ID into ip_id.c.
Correctly handle __fcntl_compat symbol for the !SYSCALL_COMPAT case.
Both .weak and .alias assembler directives only work when assembling
the file which defines the symbol.
Reported and tested by: andrew
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
A couple of internal functions used by malloc(9) and uma truncated
a size_t down to an int. This could cause any number of issues
(e.g. indefinite sleeps, memory corruption) if any kernel
subsystem tried to allocate 2GB or more through malloc. zfs would
attempt such an allocation when run on a system with 2TB or more
of RAM.
Note to self: When this is MFCed, sparc64 needs the same fix.
Differential revision: https://reviews.freebsd.org/D2106
Reviewed by: kib
Reported by: Michael Fuckner <michael@fuckner.net>
Tested by: Michael Fuckner <michael@fuckner.net>
MFC after: 2 weeks
On Ethernet packets have a minimal length, so very short packets get padding
appended to them. This padding is not stripped off in ip6_input() (due to
support for IPv6 Jumbograms, RFC2675).
That means PF needs to be careful when reassembling fragmented packets to not
include the padding in the reassembled packet.
While here also remove the 'Magic from ip_input.' bits. Splitting up and
re-joining an mbuf chain here doesn't make any sense.
Preserve IPv6 fragment IDs accross reassembly and refragmentation
When forwarding fragmented IPv6 packets and filtering with PF we
reassemble and refragment. That means we generate new fragment headers
and a new fragment ID.
We already save the fragment IDs so we can do the reassembly so it's
straightforward to apply the incoming fragment ID on the refragmented
packets.
Return EINVAL instead of EFTYPE if we have a multiboot kernel loaded but
failed to load the modules. This makes it clear that the kernel/module
should be handled by the multiboot handler but something went wrong.
Zero the list of modules array before using it, or else we might pass
uninitialized data in unused fields of the struct that will make Xen choke.
Also add a check to make sure malloc succeeds.