Recheck curthread->td_su after the VFS_SYNC() call, and re-sync if the
ast was rescheduled during VFS_SYNC(). It is possible that enough
parallel writes or slow/hung volume result in VFS_SYNC() deferring to
the ast flushing of workqueue.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Warner Losh [Mon, 21 Dec 2015 05:08:20 +0000 (05:08 +0000)]
There's no such program as [-n :(. Add the missing space. Also,
work when the object directory doesn't exist yet. realpath hates
us, so use the backup of the realpath arg when that happens.
Ian Lepore [Sun, 20 Dec 2015 23:38:14 +0000 (23:38 +0000)]
Change KERNVIRTADDR to 0xc0000000 since the low-order bits no longer need to
match the physical load address. Remove the *PHYSADDR symbols which are no
longer necessary.
Also remove a bunch of comments, most of which which have been wrong for
quite some time now, and the rest of which are mooted by these changes. All
that's left in this file is assigning a single symbol to its cannonical
value, not much to comment on anymore.
Ian Lepore [Sun, 20 Dec 2015 23:31:11 +0000 (23:31 +0000)]
Allow armv4/5 kernels to be loaded on any 2MB boundary, like armv6/7.
This eliminates the reliance on PHYSADDR and KERNPHYSADDR compile-time
symbols (except when the rom-copy code is enabled) by using the current
PC and the assumption that the entry-point routine is in the first 1MB
section of the text segment.
Other cleanups done:
- Reduce the initarm() stack size back to 2K. It got increased to
4 * 2K when this file was supporting multicore armv6, but that
support is now in locore-v6.S.
- When building the temporary startup page tables, map the entire
4GB address space as VA=PA before mapping the kernel at its loaded
location. This allows access to boot parameters stored somewhere
in ram by the bootloader, regardless of where that may be.
- When building the page table entry for supporting EARLY_PRINTF, map
the section as uncached unbuffered, since it is presumably device
registers.
Note that this restores the ability to use loader(8)/ubldr on armv4/5
kernels. That was broken in r283035, the point at which ubldr started
loading an arm kernel at any 2MB boundary.
Also note that after this, there is no reason to set KERNVIRTADDR to
anything other than 0xc0000000, and no need for PHYSADDR or KERNPHYSADDR
symbols at all.
Ian Lepore [Sun, 20 Dec 2015 19:09:12 +0000 (19:09 +0000)]
Tidy up mips ofw_machdep.h. Don't include openfirm.h because openfirm.h
is what includes machine/ofw_machdep.h. Don't declare OF_decode_addr();
it isn't implemented yet on mips and the declaration for it is about to
be commonized into openfirm.h.
Ed Maste [Sun, 20 Dec 2015 16:07:09 +0000 (16:07 +0000)]
loader.efi: refresh size in GetMemoryMap retry loop
If ExitBootServices fails due to a changed efi_mapkey then GetMemoryMap
must be called again. In this case it is also possible for the memory
map to grow, so repeat the initial GetMemoryMap call to fetch the new
size.
Also roll bi_add_efi_data_and_exit into bi_load_efi_data as there's no
need for it to be a separate function.
Depending on system configuration and parameters, clock_gettime() and
gettimeofday() may not be system calls. If so, passing an invalid pointer
will cause a signal and not an [EFAULT] error.
From a standards perspective, this is OK since passing an invalid pointer is
undefined behaviour.
Enji Cooper [Sun, 20 Dec 2015 05:52:03 +0000 (05:52 +0000)]
Always expose LOCALBASE, not just when CROSS_TOOLCHAIN is defined
Instead of using which(1) to look for doxygen, look for it in <LOCALBASE>/bin .
$PATH gets mangled by make buildenv, etc so it's better to just be explicit
about the path if someone uses that for instance.
Enji Cooper [Sun, 20 Dec 2015 04:52:53 +0000 (04:52 +0000)]
Skip the testcases on i386 (all the assertions fail) [*]
Also, don't compile the ldexpl(3) testcases on platforms that don't support
the libcall (technically only x86 right now). This makes this test buildable on
arm*/mips*/powerpc*
PR: 205449 [*]
MFC after: 1 week
Tested on: stable/10 (amd64/i386), head (amd64/i386)
Sponsored by: EMC / Isilon Storage Division
Enji Cooper [Sun, 20 Dec 2015 04:28:37 +0000 (04:28 +0000)]
- Use nitems instead of handrolling the macro
- Use a separate variable for tracking the testcase count instead
of hardcoding the offset for the testcases
Enji Cooper [Sun, 20 Dec 2015 04:17:54 +0000 (04:17 +0000)]
Disable test-ctrig.t testcases which fail assertions on i386 [*]
Also, don't compile the exp2l(3) testcases on platforms that don't support the
libcall (technically only x86 right now). This makes this test buildable on
arm*/mips*/powerpc*
Tested on: stable/10 (amd64/i386), head (amd64/i386)
PR: 205446 [*]
MFC after: 1 week
X-MFC to: stable/10
Sponsored by: EMC / Isilon Storage Division
Add a safety net to reclaim mbufs when one of the mbuf zones become
exhausted.
It is possible for a bug in the code (or, theoretically, even unusual
network conditions) to exhaust all possible mbufs or mbuf clusters.
When this occurs, things can grind to a halt fairly quickly. However,
we currently do not call mb_reclaim() unless the entire system is
experiencing a low-memory condition.
While it is best to try to prevent exhaustion of one of the mbuf zones,
it would also be useful to have a mechanism to attempt to recover from
these situations by freeing "expendable" mbufs.
This patch makes two changes:
a) The patch adds a generic API to the UMA zone allocator to set a
function that should be called when an allocation fails because the
zone limit has been reached. Because of the way this function can be
called, it really should do minimal work.
b) The patch uses this API to try to free mbufs when an allocation
fails from one of the mbuf zones because the zone limit has been
reached. The function schedules a callout to run mb_reclaim().
Add dev.fb.X.resync sysctl to resync ARM framebuffer with VideoCore
Some applications (e.g. Kodi) use tvservice APIs to manage HDMI
modes, power state, EDID etc. directly through VideoCore. After
these manipulations VideoCore may loose its state and needs to be
resynced with ARM. Under Linux this problem is worked around using
fbset utility that recreates framebuffer. Since there is no fbset
utility in FreeBSD we provide sysctl for userland apps to get system
back into normal mode.
Warner Losh [Sat, 19 Dec 2015 19:13:38 +0000 (19:13 +0000)]
Create libstand version of uuid_from_string and uuid_to_string. The
normal libc versions of these functions use asprintf and sscanf, both
of which aren't available in the libstand world. It's easier to
rewrite these than to try to claw those functions into libstand.
Also include uuid_create_nil, since it's safe to include. These were
written from scratch, except for bits of comments that I got from the
libc version, which are acknowledged in the source.
Warner Losh [Sat, 19 Dec 2015 19:01:42 +0000 (19:01 +0000)]
Pass the UEFI system table into the kernel from the boot loader via
the FW_HANDLE metadata item. The kernel can get the rest of what it
needs from this.
Alan Cox [Sat, 19 Dec 2015 18:42:50 +0000 (18:42 +0000)]
Introduce a new mechanism for relocating virtual pages to a new physical
address and use this mechanism when:
1. kmem_alloc_{attr,contig}() can't find suitable free pages in the physical
memory allocator's free page lists. This replaces the long-standing
approach of scanning the inactive and inactive queues, converting clean
pages into PG_CACHED pages and laundering dirty pages. In contrast, the
new mechanism does not use PG_CACHED pages nor does it trigger a large
number of I/O operations.
2. on 32-bit MIPS processors, uma_small_alloc() and the pmap can't find
free pages in the physical memory allocator's free page lists that are
covered by the direct map. Tested by: adrian
3. ttm_bo_global_init() and ttm_vm_page_alloc_dma32() can't find suitable
free pages in the physical memory allocator's free page lists.
In the coming months, I expect that this new mechanism will be applied in
other places. For example, balloon drivers should use relocation to
minimize fragmentation of the guest physical address space.
Make vm_phys_alloc_contig() a little smarter (and more efficient in some
cases). Specifically, use vm_phys_segs[] earlier to avoid scanning free
page lists that can't possibly contain suitable pages.
Nathan Whitehorn [Sat, 19 Dec 2015 02:16:38 +0000 (02:16 +0000)]
Provide link state reporting so that ifconfig_llan0="DHCP" works. The
reported link state is fictional (always up) since the hypervisor does
not provide this information.
Ed Maste [Fri, 18 Dec 2015 17:39:54 +0000 (17:39 +0000)]
loader.efi: show EFI error number, not full status value
EFI return values set the high bit to indicate an error. The log
messages changed here are printed only in the case of an error,
so including the error bit is redundant. Also switch to decimal to
match the error definitions (in sys/boot/efi/include/efierr.h).
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
Hajimu UMEMOTO [Fri, 18 Dec 2015 17:30:22 +0000 (17:30 +0000)]
Generate ipnodes.by{addr,name} from /etc/hosts for compatibility with
FreeBSD local name resolution.
If /var/yp/ipnodes exists, we generate them from it for backward
compatibility.
Mateusz Guzik [Fri, 18 Dec 2015 16:33:15 +0000 (16:33 +0000)]
proc: fix a race which could result in dereference of bad p_pgrp pointer on fork
During fork p_starcopy - p_endcopy area of a process is populated with bcopy
with only proc lock held. Another forking thread can find such a process and
proceed to access p_pgrp included in said area.
Fix the problem by moving the field outside. It is being properly assigned
later.
Nathan Whitehorn [Fri, 18 Dec 2015 14:35:27 +0000 (14:35 +0000)]
Enable PRINTF_BUFR_SIZE on powerpc64, following r194204 on x86. The kernel
message garbling was becoming very noticeable on the 64-CPU systems we now
support and run on.
Warner Losh [Fri, 18 Dec 2015 06:13:18 +0000 (06:13 +0000)]
Add a generic firmware dependent handle to pass from the loader to the
kernel. This will be used for passing in things like the system table
from EFI or other similar metadata that can be used by the kernel to
communicate with the firmware.
Adrian Chadd [Fri, 18 Dec 2015 05:43:59 +0000 (05:43 +0000)]
[intrng] Migrate the intrng code from sys/arm/arm to sys/kern/subr_intr.c.
The ci20 port (by kan@) is going to reuse almost all of the intrng code
since the SoC in question looks suspiciously like someone took an ARM
SoC design and replaced the ARM core with a MIPS core.
* migrate out the code;
* rename ARM_ -> INTR_;
* rename arm_ -> intr_;
* move the interrupt flush routine from intr.c / intrng.c into
arm/machdep_intr.c - removing the code duplication and removing
the ARM specific bits from here.
Thanks to the Star Wars: The Force Awakens premiere line for allowing
me a couple hours of quiet time to finish the universe builds.
Tested:
* make universe
TODO:
* The structure definitions in subr_intr.c still includes machine/intr.h
which requires one duplicates all of the intrng definitions in
the platform code (which kan has done, and I think we don't have to.)
Instead I should break out the generic things (function declarations,
common intr structures, etc) into a separate header.
* Kan has requested I make the PIC based IPI stuff optional.
Ian Lepore [Fri, 18 Dec 2015 01:25:30 +0000 (01:25 +0000)]
Fix the clock divisor calc for imx6 sdcard bus speed.
I don't know what alternate universe I was inhabiting when I wrote it
originally, but apparently the basic workings of mathematics were different
than in this universe. I also can't explain how it ever worked, except "by
accident", because completely bogus values were being written into the
divisor register.
John Baldwin [Thu, 17 Dec 2015 20:33:20 +0000 (20:33 +0000)]
t seems certain Intel GPUs use GPIO bitbanging over a child device
instead of GMBUS access for I2C transfers. The GMBUS driver falls back
to this mode when a transfer times out. However, the first transfer to
timeout was sending the request back to itself resulting in an panic due
to recursing on a lock. Fix it to forward the request on to the proper
device. This appears to have been accidentally changed in r277487.
Conrad Meyer [Thu, 17 Dec 2015 17:48:57 +0000 (17:48 +0000)]
vm_page_replace: add wrapper to KASSERT about old page
It turns out the callers of vm_page_replace know exactly which page they are
replacing and would like to assert about it. Change those from hard panics to
KASSERTs, and provide them with a wrapper so they don't have to deal with
warnings from an INVARIANTS-dependent dead store of the return value of
vm_page_replace.
Andrew Turner [Thu, 17 Dec 2015 17:00:04 +0000 (17:00 +0000)]
Support the variant of the interrupt-map property where the parent bus has
the #address-cells property set. For this we need to read more data before
the parent interrupt description.
this is only enabled on arm64 for now as it's not quite compliant with the
ePAPR spec. We should use a default of 2 where the #address-cells property
is missing, however this will need further testing across architectures.
Switch the IPsec related statistics to using the built in sysctl
variable set rather than reading from kernel memory.
This also makes the -z (zero) flag work correctly
Mark Johnston [Thu, 17 Dec 2015 00:00:27 +0000 (00:00 +0000)]
Support an arbitrary number of arguments to DTrace syscall probes.
Rather than pushing all eight possible arguments into dtrace_probe()'s
stack frame, make the syscall_args struct for the current syscall available
via the current thread. Using a custom getargval method for the systrace
provider, this allows any syscall argument to be fetched, even in kernels
that have modified the maximum number of system call arguments.