Also return error if TSO is requested without Tx checksum offload.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2564
It just affects capabilities of the created VLAN interface.
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2563
If TxQ lock is obtained, deferred packet list shold be serviced even if
the packet addition fails because of overflow.
Without the patch freeze happens if:
- queue is not blocked (i.e. completion does not trigger unblock and service)
- put-list overflow (1024 entries)
- sfxge_tx_packet_add() acquires TxQ lock just as it is released it in
sfxge_tx_qdpl_service() on the second CPU but before pending check
- sfxge_tx_packet_add() swizzles put-list to get-list, fails because of
non-tcp get-list overflow and returns without packet list service
- sfxge_tx_qdpl_service() on the second CPU checks that there are no
pending packets in the put-list and returns
Other possible solution is to guaranee that maximum length of the put-list
is less than maximum length of any get-list.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2562
Previously, ubldr would use the virtual addresses in the elf headers by
masking off the high bits and assuming the result was a physical address
where the kernel should be loaded. That would sometimes discard
significant bits of the physical address, but the effects of that were
undone by archsw copy code that would find a large block of memory and
apply an offset to the source/dest copy addresses. The result was that
things were loaded at a different physical address than requested by the
higher code layers, but that worked because other adjustments were applied
later (such as when jumping to the entry point). Very confusing, and
somewhat fragile.
Now the archsw copy routines are just simple copies, and instead
archsw.arch_loadaddr is implemented to choose a load address. The new
routine uses some of the code from the old offset-translation routine to
find the largest block of ram, but it excludes ubldr itself from that
range, and also excludes If ubldr splits the largest block of ram in
two, the kernel is loaded into the bottom of whichever resulting block is
larger.
As part of eliminating ubldr itself from the ram ranges, export the heap
start/end addresses in a pair of new global variables.
This change means that the virtual addresses in the arm kernel elf headers
now have no meaning at all, except for the entry point address. There is
an implicit assumption that the entry point is in the first text page, and
that the address in the the header can be turned into an offset by masking
it with PAGE_MASK. In the future we can link all arm kernels at a virtual
address of 0xC0000000 with no need to use any low-order part of the
address to influence where in ram the kernel gets loaded.
1. Align to a 64-bit address so 64-bit data will be correctly aligned.
2. Add a comment explaining why.
3. Remove an unneeded value from the struct.
This fixes an issue where the struct may not be correctly aligned on the
stack in the syscall function. This may lead to accesing a 64-bit value
at a non 64-bit. This will raise an exception and panic the kernel.
We have been lucky where on arm and armv6 both clang and gcc correctly
align the data, even without us asking to, however, on armeb with clang to
not be the case. This tells the compiler we really do need this to be
aligned.
Reported and tested by: jmg (on armeb with clang)
MFC after: 1 Week [1, 2]
when loader(8) passed physical addresses in loader metadata for arm, but
that is no longer true; all metadata has already been adjusted to vitual
addresses by loader.
I can't track down the exact revision in loader where a change from physical
to virtual metadata addresses happened. The code involved is very twisty
and complicated. I suspect the change was an unintended consequence of the
r247301, r247413, r248118 series of changes I made a couple years ago.
comment to this effect and switch the default. My old AT91SAM9G20
now boots, fsck's the SD card and runs w/o an issue for the first
time since a 9.1-ish stable build I did a few years ago.
Problems with unmapped I/O:
o un-page-aligned I/O requests to devices fail (notably fsck
and newfs).
o write-back caching was totally broken. write-through caching
needed to be enabled.
o Even page-aligned I/O requests sometimes failed for reasons
not thoroughly investigated.
Suggested by: ian@
MFC after: 2 days
The driver uses ifm_data to save capabilities mask calculated during
initialization when supported phy modes are discovered.
The patch simply calculates it when either media or options are changed.
Reviewed by: glebius
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2540
It is required for the next patch which adds dependency of TSO
capabilities from Tx checksum offloads.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2553
* simplify channel logic for determining RF gain setting in scan setup
* don't set TX timer on error
* free node references for unsent frames on device stop
* set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command)
* add missing IWN_UNLOCK() from interrupt path when the hardware
disappears.
* pass control frames to host
* nitems() instead of local macro
Tested:
* Intel 5100, STA mode
PR: kern/196264
Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
It is a preparation to the next patch which will service packet queue even
if packet addtion fails.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2552
Now each branch has one and only one possible TxQ lock state.
It simplifies understanding of the code.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2551
We can't disable it in HW, but we can ignore result.
Discard Rx descriptor checksum flags if Rx checksum offload is off.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2544
This lets the compiler know about the alignment of pointers returned
by aligned_alloc(3), posix_memalign(3). and contigmalloc(9)
Currently this is only supported in recent gcc but we are ready to
use it if clang implements it.
Relnotes: yes
In general it is bad practice to use the gnu_inline attribute but we
will need it in special cases like FORTIFY_SOURCE. In this specific
case it is also useful to have the "artificial" attribute:
"This attribute is useful for small inline wrappers which if possible
should appear during debugging as a unit, depending on the debug info
format it will either mean marking the function as artificial or using the
caller location for all instructions within the inlined body."
This attribute appears to be currently implemented only in GCC. Use it
only in conjuntion with gnu_inline in the cases where it is available,
which is similar in spirit in how it's used in glibc.
The Alpine Platform-On-Chip offers multicore processing
(quad ARM Cortex-A15), 1/10Gb Ethernet, SATA 3, PCI-E 3,
DMA engines, Virtualization, Advanced Power Management and other.
This code drop involves basic platform support including:
SMP, IRQs, SerDes, SATA. As of now it is missing the PCIe support.
Part of the functionality is provided by the low-level code (HAL)
delivered by the chip vendor (Annapurna Labs) and is a subject to
change in the future (is planned to be moved to sys/contrib directory).
The review log for this commit is available here:
https://reviews.freebsd.org/D2340
Reviewed by: andrew, ian, imp
Obtained from: Semihalf
Sponsored by: Annapurna Labs
Perform cache writebacks and invalidations in the correct (inner to outer
or vice versa) order, and add comments that explain that.
Consistantly use 'va' as the variable name for virtual addresses.
Submitted by: Michal Meloun <meloun@miracle.cz>
was granted via rings and ni_bufs_list_head represented in those rings
and lists (e.g., via SIGKILL), those buffers are no longer available
for subsequent users for the lifetime of the system. To mitigate this
resource leak, reset the allocator state when the last ref to that
allocator is released.
Note that this only recovers leaked resources for an allocator when
there are no longer any users of that allocator, so there remain
circumstances in which leaked allocator resources may not ever be
recovered - consider a set of multiple netmap processes that are all
using the same allocator (say, the global allocator) where members of
that set may be killed and restarted over time but at any given point
there is one member of that set running.
Based on intial work by adrian@.
Reviewed by: Giuseppe Lettieri (g.lettieri@iet.unipi.it), luigi
Approved by: jmallett (mentor)
MFC after: 1 week
Sponsored by: Norse Corp, Inc.
thread awakened due to a time out, then cv_waiters was not decremented.
If INT_MAX threads timed out on a cv without an intervening cv_broadcast,
then cv_waiters could overflow. To fix this, have each sleeping thread
decrement cv_waiters when it resumes.
Note that previously cv_waiters was protected by the sleepq chain lock.
However, that lock is not held when threads resume from sleep. In
addition, the interlock is also not always reacquired after resuming
(cv_wait_unlock), nor is it always held by callers of cv_signal() or
cv_broadcast(). Instead, use atomic ops to update cv_waiters. Since
the sleepq chain lock is still held on every increment, it should
still be safe to compare cv_waiters against zero while holding the
lock in the wakeup routines as the only way the race should be lost
would result in extra calls to sleepq_signal() or sleepq_broadcast().
Differential Revision: https://reviews.freebsd.org/D2427
Reviewed by: benno
Reported by: benno (wrap of cv_waiters in the field)
MFC after: 2 weeks
gif(4) interface. Add new option "ignore_source" for gif(4) interface.
When it is enabled, gif's encapcheck function requires match only for
packet's destination address.
Differential Revision: https://reviews.freebsd.org/D2004
Obtained from: Yandex LLC
MFC after: 2 weeks
Sponsored by: Yandex LLC
particular, switch to the proc0 pmap to have expected %cr3 and PCID
for the thread0 during initialization, and the up to date pm_active
mask.
pmap_pinit0() should be done after proc0->p_vmspace is assigned so
that the amd64 pmap_activate() find the correct curproc pmap.
Sponsored by: The FreeBSD Foundation
MFC after: 3 weeks
suspended thread itself, on the return path from
thread_suspend_check(). A consequence is that return from
thread_single_end(SINGLE_BOUNDARY) may leave p_boundary_count
non-zero, it might be even equal to the threads count.
Now, assume that we have two threads in the process, both calling
execve(2). Suppose that the first thread won the race to be the
suspension thread, and that afterward its exec failed for any reason.
After the first thread did thread_single_end(SINGLE_BOUNDARY), second
thread becomes the process suspension thread and checks
p_boundary_count. The non-zero value of the count allows the
suspension loop to finish without actually suspending some threads.
In other words, we enter exec code with some threads not suspended.
Fix this by decrementing p_boundary_count in the
thread_single_end()->thread_unsuspend_one() during marking the thread
as runnable. This way, a return from thread_single_end() guarantees
that the counter is cleared. We do not care whether the unsuspended
thread has a chance to run.
Add some asserts to ensure the state of the process when single
boundary suspension is lifted. Also make thread_unuspend_one()
static.
In collaboration with: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
It simplifies understanding of the sfxge_tx_packet_add() logic and
avoids passing of 'locked' to called function.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2547
It is simply not required since the kernel checks corresponding
IFCAP_TSOx capability and CSUM_TSO in hw-assisted offloads.
Note that CSUM_TSO is two bits (CSUM_IP_TSO|CSUM_IP6_TSO) and both bits
are set in IPv4 and IPv6 mbufs.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2546
Also it is cheaper to check Rx descriptor flags than TCP protocol in IP
header.
Reviewed by: gnn
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2542
For consistency with the naming conventions used by the other
implementations kill armv7_sleep and keep armv7_cpu_sleep.
Differential Revision: https://reviews.freebsd.org/D2537
Submitted by: John Wehle
Reviewed by: ian@, andrew@
Netfront has to wait for the backend to switch to state XenbusStateConnected
before sending the ARP request, or else the backend might not be connected
and thus the packet will be lost.
Sponsored by: Citrix Systems R&D
MFC after: 1 week
This is one of the few post gcc4.2 builtins that has been implemented by
clang:
__builtin_unreachable is used to indicate that a specific point in the
program cannot be reached, even if the compiler might otherwise think it
can. This is useful to improve optimization and eliminates certain
warnings.
Hinted by: NetBSD
Differential Revision: https://reviews.freebsd.org/D2536
This allows functions that retrieve and inspect pthread_attr_t objects to
work correctly: querying the cpuset_t size is part of querying CPU
affinity information, which is part of creating a complete pthread_attr_t.
Approved by: rwatson (mentor)
Reviewed by: pjd
Sponsored by: NSERC
Tx checksum offload may be enabled/disabled.
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D2543
