[HEXAGON] [MachinePipeliner] Fix the DAG in case of dependent phis. (#135925)
This change corrects the scheduling relationship between dependent PHI
nodes. Previously, the implementation treated SU1 as the successor of
SU0. In reality, SU0 should depend on SU1, not the other way around.
The incorrect ordering could cause SU0 to be scheduled before SU1, which
leads to invalid IR: subsequent instructions may reference values that
have not yet been defined.
%3:intregs = PHI %21:intregs, %bb.6, %7:intregs, %bb.1 - SU0 %7:intregs
= PHI %21:intregs, %bb.6, %13:intregs, %bb.1 - SU1 %27:intregs = A2_zxtb
%3:intregs - SU2
%13:intregs = C2_muxri %45:predregs, 0, %46:intreg
Co-Authored by: Sumanth Gundapaneni
HBSD: Fix build for devel/omniORB
While here, enable PIE support. Its applications build fine as PIEs.
Signed-off-by: Shawn Webb <shawn.webb at hardenedbsd.org>
Merge tag 'spi-v6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Pull spi updates from Mark Brown:
"This release is almost entirely new drivers, with a couple of small
changes in generic code.
The biggest individual update is a rename of the existing Microchip
driver and the addition of a new driver for the silicon SPI controller
in their PolarFire SoCs. The overlap between the soft IP supported by
the current driver and this new one is regrettably all in the IP and
not in the register interface offered to software.
- Add a time offset parameter for offloads, allowing them to be
defined in relation to each other. This is useful for IIO type
applcations where you trigger an operation then read the result
after a delay.
- Add a tracepoint for flash exec_ops, bringing the flash support
more in line with the debuggability of vanilla SPI.
[32 lines not shown]
Merge tag 'regulator-v6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator
Pull regulator updates from Mark Brown:
"This is a relatively busy release for the regulator API, as well as a
good collection of new drivers we've got a little bit of core work and
a bunch of cleanup throughout the subsystem:
- Support for propagating undervoltage events to child regulators
- Undo enables done on supplies when setting enabling regulators via
constraints fails
- Pull in some gpiolib changes adding support for shared GPIOs to the
gpiolib core, using them to replace the open coded variant of this
that we've had in the regulator API for a long time
- Support for Fitipower FP9931 and JD9330, Mediatek MT6316, MT6363
and MT6373, NXP PF1550 and Qualcomm PMH01XX and PMCX0102
[26 lines not shown]
Merge tag 'regmap-v6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap
Pull regmap updates from Mark Brown:
"Another small update for regmap, we have one new feature plus a little
bit of cleanup:
- Support for sparseness information in the flat cache, allowing
users that really need the performance properties it provides to
benefit from the interface and startup time improvements that
sparsness provides without needing to go all the way to a more
fancy data structure
- Cleanup work from Andy Shevchenko, refactoring the cache interface
in preparation for some future stuff he's working on"
* tag 'regmap-v6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
regmap: sdw-mbq: Reorder regmap_mbq_context struct for better packing
regmap: i3c: Use ARRAY_SIZE()
regcache: maple: Split ->populate() from ->init()
[6 lines not shown]
net: remove dom_ifmtu
It is a remnant of a network stack design that was supposed to support
multiple network protocols. Today it is clear that we are left with IPv4
and IPv6 only. Only IPv6 may have an MTU different to the interface MTU.
Revert [VPlan] Consolidate logic for narrowToSingleScalars (#170720)
This reverts commit 7b3ec51, as a crash was reported:
https://llvm.godbolt.org/z/dK6ff5zvr -- this will give us time to
investigate a re-land.
[SystemZ] Handle IR struct arguments correctly. (#169583)
- The size of the stack slot was previously computed in LowerCall() by using
the original type, but that didn't work for a struct. Compute the size
by looking at the VT of each part and the number of them instead.
- All the members of a struct have the same OrigArgIndex, so it doesn't work
to assume that following parts belong to a split argument until another
OrigArgIndex is encountered. Use the isSplit() and isSplitEnd() flags
instead.
- Detect any scalar integer argumet >64 bits in CanLowerReturn() instead of
just i128, in order to let all of them be passed on stack.
Fixes #168460
Merge tag 'mtd/for-6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux
Pull mtd updates from Miquel Raynal:
"Core MTD changes:
- We must ignore error -ENOENT from parsers on subpartitions which is
a legitimate return value
- PM support is added to the intel-dg driver
Raw NAND changes:
- The major change in this is the support for the Allwinner H616 NAND
controller, which lead to numerous changes and cleanups in the
driver.
- Another notable change in this driver is the use of field_get() and
field_prep(), but since the global support for this helpers is
going to be merged in the same release as we start using these
[58 lines not shown]
Merge tag 'pwm/for-6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/ukleinek/linux
Pull pwm updates from Uwe Kleine-König:
"In addition to the usual mix of core cleanups, driver changes, minor
fixes and device tree updates the highlight this cycle is Rust support
for the core and a first Rust driver both provided by Michal
Wilczynski.
Michal wrote about these changes on
https://mwilczynski.dev/posts/bringing-rust-to-the-pwm-subsystem/
which is a nice read"
* tag 'pwm/for-6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/ukleinek/linux: (22 commits)
pwm: rzg2l-gpt: Allow checking period_tick cache value only if sibling channel is enabled
pwm: bcm2835: Make sure the channel is enabled after pwm_request()
pwm: mediatek: Make use of struct_size macro
pwm: mediatek: Remove unneeded semicolon
[17 lines not shown]
[OpenACC][CIR] Implement 'gang' lowering on `routine' (#170506)
This is a bit more work than the worker/vector/seq in that gang takes an
optional `dim` argument. The argument is always 1, 2, or 3 (constants!),
and the other argument-types that gang allows elsewhere aren't valid
here.
For the IR, we had to add 2 overloads of `addGang`. The first just adds
the 'valueless' one, which can just add to the one ArrayAttr. The second
has to add to TWO lists.
Note: The standard limits to only 1 `gang` per construct. We decided
after evaluating it, that it really means 'per device-type region'.
However, device_type isn't implemented yet, so we'll add tests for that
when we do.
At the moment, we added the device_type infrastructure however.
