- 25 Dec, 2017 40 commits
-
-
Daniel Borkmann authored
[ Upstream commit 07aee943 ] When LD_ABS/IND is used in the program, and we have a BPF helper call that changes packet data (bpf_helper_changes_pkt_data() returns true), then in case of sparc JIT, we try to reload cached skb data from bpf2sparc[BPF_REG_6]. However, there is no such guarantee or assumption that skb sits in R6 at this point, all helpers changing skb data only have a guarantee that skb sits in R1. Therefore, store BPF R1 in L7 temporarily and after procedure call use L7 to reload cached skb data. skb sitting in R6 is only true at the time when LD_ABS/IND is executed. Fixes: 7a12b503 ("sparc64: Add eBPF JIT.") Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
David S. Miller <davem@davemloft.net> Acked-by:
Alexei Starovoitov <ast@kernel.org> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Daniel Borkmann authored
[ Upstream commit 87338c8e ] The assumption of unconditionally reloading skb pointers on BPF helper calls where bpf_helper_changes_pkt_data() holds true is wrong. There can be different contexts where the helper would enforce a reload such as in case of XDP. Here, we do have a struct xdp_buff instead of struct sk_buff as context, thus this will access garbage. JITs only ever need to deal with cached skb pointer reload when ld_abs/ind was seen, therefore guard the reload behind SEEN_SKB. Fixes: 156d0e29 ("powerpc/ebpf/jit: Implement JIT compiler for extended BPF") Signed-off-by:
Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com> Acked-by:
Sandipan Das <sandipan@linux.vnet.ibm.com> Signed-off-by:
-
Daniel Borkmann authored
[ Upstream commit 6d59b7db ] The assumption of unconditionally reloading skb pointers on BPF helper calls where bpf_helper_changes_pkt_data() holds true is wrong. There can be different contexts where the BPF helper would enforce a reload such as in case of XDP. Here, we do have a struct xdp_buff instead of struct sk_buff as context, thus this will access garbage. JITs only ever need to deal with cached skb pointer reload when ld_abs/ind was seen, therefore guard the reload behind SEEN_SKB only. Tested on s390x. Fixes: 9db7f2b8 ("s390/bpf: recache skb->data/hlen for skb_vlan_push/pop") Signed-off-by:
-
Daniel Borkmann authored
[ Upstream commit 283ca526 ] When tracing and networking programs are both attached in the system and both use event-output helpers that eventually call into perf_event_output(), then we could end up in a situation where the tracing attached program runs in user context while a cls_bpf program is triggered on that same CPU out of softirq context. Since both rely on the same per-cpu perf_sample_data, we could potentially corrupt it. This can only ever happen in a combination of the two types; all tracing programs use a bpf_prog_active counter to bail out in case a program is already running on that CPU out of a different context. XDP and cls_bpf programs by themselves don't have this issue as they run in the same context only. Therefore, split both perf_sample_data so they cannot be accessed from each other. Fixes: 20b9d7ac ("bpf: avoid excessive stack usage for perf_sample_data") Reported-by:
Alexei Starovoitov <ast@fb.com> Signed-off-by:
Song Liu <songliubraving@fb.com> Acked-by:
-
Daniel Borkmann authored
From: Alexei Starovoitov <ast@fb.com> [ Upstream commit c131187d ] when the verifier detects that register contains a runtime constant and it's compared with another constant it will prune exploration of the branch that is guaranteed not to be taken at runtime. This is all correct, but malicious program may be constructed in such a way that it always has a constant comparison and the other branch is never taken under any conditions. In this case such path through the program will not be explored by the verifier. It won't be taken at run-time either, but since all instructions are JITed the malicious program may cause JITs to complain about using reserved fields, etc. To fix the issue we have to track the instructions explored by the verifier and sanitize instructions that are dead at run time with NOPs. We cannot reject such dead code, since llvm generates it for valid C code, since it doesn't do as much data flow analysis as the verifier does. Fixes: 17a52670 ("bpf: verifier (add verifier core)") Signed-off-by:
-
Kirill A. Shutemov authored
commit 629a359b upstream. Since commit: 83e3c487 ("mm/sparsemem: Allocate mem_section at runtime for CONFIG_SPARSEMEM_EXTREME=y") we allocate the mem_section array dynamically in sparse_memory_present_with_active_regions(), but some architectures, like arm64, don't call the routine to initialize sparsemem. Let's move the initialization into memory_present() it should cover all architectures. Reported-and-tested-by:
Sudeep Holla <sudeep.holla@arm.com> Tested-by:
Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by:
Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by:
Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mm@kvack.org Fixes: 83e3c487 ("mm/sparsemem: Allocate mem_section at runtime for CONFIG_SPARSEMEM_EXTREME=y") Link: http://lkml.kernel.org/r/20171107083337.89952-1-kirill.shutemov@linux.intel.comSigned-off-by:
Ingo Molnar <mingo@kernel.org> Cc: Dan Rue <dan.rue@linaro.org> Cc: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by:
-
Peter Hutterer authored
commit bff5bf9d upstream. Sending the switch state change twice within the same frame is invalid evdev protocol and only works if the client handles keys immediately as well. Processing events immediately is incorrect, it forces a fake order of events that does not exist on the device. Recent versions of libinput changed to only process the device state and SYN_REPORT time, so now the key event is lost. https://bugs.freedesktop.org/show_bug.cgi?id=104041Signed-off-by:
Peter Hutterer <peter.hutterer@who-t.net> Signed-off-by:
Darren Hart (VMware) <dvhart@infradead.org> Signed-off-by:
-
Daniel Lezcano authored
commit db2b0332 upstream. The DT specifies a threshold of 65000, we setup the register with a value in the temperature resolution for the controller, 64656. When we reach 64656, the interrupt fires, the interrupt is disabled. Then the irq thread runs and calls thermal_zone_device_update() which will call in turn hisi_thermal_get_temp(). The function will look if the temperature decreased, assuming it was more than 65000, but that is not the case because the current temperature is 64656 (because of the rounding when setting the threshold). This condition being true, we re-enable the interrupt which fires immediately after exiting the irq thread. That happens again and again until the temperature goes to more than 65000. Potentially, there is here an interrupt storm if the temperature stabilizes at this temperature. A very unlikely case but possible. In any case, it does not make sense to handle dozens of alarm interrupt for nothing. Fix this by rounding the threshold value to the controller resolution so the check against the threshold is consistent with the one set in the controller. Signed-off-by:
Daniel Lezcano <daniel.lezcano@linaro.org> Reviewed-by:
Leo Yan <leo.yan@linaro.org> Tested-by:
Eduardo Valentin <edubezval@gmail.com> Signed-off-by:
Kevin Wangtao <kevin.wangtao@hisilicon.com> Signed-off-by:
-
Daniel Lezcano authored
commit 48880b97 upstream. The step and the base temperature are fixed values, we can simplify the computation by converting the base temperature to milli celsius and use a pre-computed step value. That saves us a lot of mult + div for nothing at runtime. Take also the opportunity to change the function names to be consistent with the rest of the code. Signed-off-by:
-
Daniel Lezcano authored
commit 2cb4de78 upstream. The threaded interrupt for the alarm interrupt is requested before the temperature controller is setup. This one can fire an interrupt immediately leading to a kernel panic as the sensor data is not initialized. In order to prevent that, move the threaded irq after the Tsensor is setup. Signed-off-by:
-
Daniel Lezcano authored
commit c176b10b upstream. The interrupt for the temperature threshold is not enabled at the end of the probe function, enable it after the setup is complete. On the other side, the irq_enabled is not correctly set as we are checking if the interrupt is masked where 'yes' means irq_enabled=false. irq_get_irqchip_state(data->irq, IRQCHIP_STATE_MASKED, &data->irq_enabled); As we are always enabling the interrupt, it is pointless to check if the interrupt is masked or not, just set irq_enabled to 'true'. Signed-off-by:
-
Niranjana Vishwanathapura authored
[ Upstream commit b77eb45e ] Do not include EM specified MAC address in total MACs of the UC MAC list. Reviewed-by:
Sudeep Dutt <sudeep.dutt@intel.com> Signed-off-by:
Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com> Signed-off-by:
Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by:
Doug Ledford <dledford@redhat.com> Signed-off-by:
Sasha Levin <alexander.levin@verizon.com> Signed-off-by:
-
Scott Franco authored
[ Upstream commit 4bbdfe25 ] Clear the MAC table digest when the MAC table is freed. Reviewed-by:
Scott Franco <safranco@intel.com> Signed-off-by:
-
Eric Anholt authored
[ Upstream commit af2eca53 ] The incoming mode might have a missing vrefresh field if it came from drmModeSetCrtc(), which the kernel is supposed to calculate using drm_mode_vrefresh(). We could either use that or the adjusted_mode's original vrefresh value. However, we can maintain a more exact vrefresh value (not just the integer approximation), by scaling by the ratio of our clocks. v2: Use math suggested by Andrzej Hajda instead. v3: Simplify math now that adjusted_mode->clock isn't padded. v4: Drop some parens. Signed-off-by:
Eric Anholt <eric@anholt.net> Link: https://patchwork.freedesktop.org/patch/msgid/20170815234722.20700-2-eric@anholt.netReviewed-by:
Andrzej Hajda <a.hajda@samsung.com> Signed-off-by:
-
Nicholas Piggin authored
[ Upstream commit f187851b ] When failing to enter broadcast timer mode for an idle state that requires it, a new state is selected that does not require broadcast, but the broadcast variable remains set. This causes tick_broadcast_exit to be called despite not having entered broadcast mode. This causes the WARN_ON_ONCE(!irqs_disabled()) to trigger in some cases. It does not appear to cause problems for code today, but seems to violate the interface so should be fixed. Signed-off-by:
Nicholas Piggin <npiggin@gmail.com> Reviewed-by:
Thomas Gleixner <tglx@linutronix.de> Signed-off-by:
Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by:
-
Alexandre Belloni authored
[ Upstream commit 74717b28 ] If there is any non expired timer in the queue, the RTC alarm is never set. This is an issue when adding a timer that expires before the next non expired timer. Ensure the RTC alarm is set in that case. Fixes: 2b2f5ff0 ("rtc: interface: ignore expired timers when enqueuing new timers") Signed-off-by:
Alexandre Belloni <alexandre.belloni@free-electrons.com> Signed-off-by:
-
Hoang Tran authored
[ Upstream commit cf5d74b8 ] With the commit 76174004 (tcp: do not slow start when cwnd equals ssthresh), the comparison to the reduced cwnd in tcp_vegas_ssthresh() would under-evaluate the ssthresh. Signed-off-by:
Hoang Tran <hoang.tran@uclouvain.be> Signed-off-by:
-
Chen-Yu Tsai authored
[ Upstream commit 7f3ed791 ] The HDMI DDC clock found in the CCU is the parent of the actual DDC clock within the HDMI controller. That clock is also named "hdmi-ddc". Rename the one in the CCU to "ddc". This makes more sense than renaming the one in the HDMI controller to something else. Fixes: c6e6c96d ("clk: sunxi-ng: Add A31/A31s clocks") Signed-off-by:
Chen-Yu Tsai <wens@csie.org> Signed-off-by:
Maxime Ripard <maxime.ripard@free-electrons.com> Signed-off-by:
-
Arvind Yadav authored
[ Upstream commit 04820da2 ] Free memory region, if gb_lights_channel_config is not successful. Signed-off-by:
Arvind Yadav <arvind.yadav.cs@gmail.com> Reviewed-by:
Rui Miguel Silva <rmfrfs@gmail.com> Signed-off-by:
-
Wei Hu(Xavier) authored
[ Upstream commit 5e437b1d ] After the loop in hns_roce_v1_mr_free_work_fn function, it is possible that all qps will have been freed (in which case ne will be 0). If that happens, then later in the function when we dereference hr_qp we will get an exception. Check ne is not 0 to make sure we actually have an hr_qp left to work on. This patch fixes the smatch error as below: drivers/infiniband/hw/hns/hns_roce_hw_v1.c:1009 hns_roce_v1_mr_free_work_fn() error: we previously assumed 'hr_qp' could be null Signed-off-by:
Wei Hu (Xavier) <xavier.huwei@huawei.com> Signed-off-by:
Lijun Ou <oulijun@huawei.com> Signed-off-by:
Shaobo Xu <xushaobo2@huawei.com> Signed-off-by:
-
Mike Manning authored
[ Upstream commit 1f372c7b ] The NS for DAD are sent on admin up as long as a valid qdisc is found. A race condition exists by which these packets will not egress the interface if the operational state of the lower device is not yet up. The solution is to delay DAD until the link is operationally up according to RFC2863. Rather than only doing this, follow the existing code checks by deferring IPv6 device initialization altogether. The fix allows DAD on devices like tunnels that are controlled by userspace control plane. The fix has no impact on regular deployments, but means that there is no IPv6 connectivity until the port has been opened in the case of port-based network access control, which should be desirable. Signed-off-by:
Mike Manning <mmanning@brocade.com> Signed-off-by:
-
Mick Tarsel authored
[ Upstream commit e876a8a7 ] State is initially reported as UNKNOWN. Before register call netif_carrier_off(). Once the device is opened, call netif_carrier_on() in order to set the state to UP. Signed-off-by:
Mick Tarsel <mjtarsel@linux.vnet.ibm.com> Signed-off-by:
-
Jacob Keller authored
[ Upstream commit 17a91809 ] When we process VF mailboxes, the driver is likely going to also queue up messages to the switch manager. This process merely queues up the FIFO, but doesn't actually begin the transmission process. Because we hold the mailbox lock during this VF processing, the PF<->SM mailbox is not getting processed at this time. Ensure that we actually process the PF<->SM mailbox in between each PF<->VF mailbox. This should ensure prompt transmission of the messages queued up after each VF message is received and handled. Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Tested-by:
Krishneil Singh <krishneil.k.singh@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
-
Marek Szyprowski authored
[ Upstream commit a99897f5 ] Odroid HC1 board has built-in JMicron USB to SATA bridge, which supports UAS protocol. Compile-in support for it (instead of enabling it as module) to make sure that all built-in storage devices are available for rootfs. The bridge itself also supports fallback to standard USB Mass Storage protocol, but USB Mass Storage class doesn't bind to it when UAS is compiled as module and modules are not (yet) available. Signed-off-by:
Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by:
Krzysztof Kozlowski <krzk@kernel.org> Signed-off-by:
-
Alex Williamson authored
[ Upstream commit 52318497 ] With virtual PCI-Express chipsets, we now see userspace/guest drivers trying to match the physical MPS setting to a virtual downstream port. Of course a lone physical device surrounded by virtual interconnects cannot make a correct decision for a proper MPS setting. Instead, let's virtualize the MPS control register so that writes through to hardware are disallowed. Userspace drivers like QEMU assume they can write anything to the device and we'll filter out anything dangerous. Since mismatched MPS can lead to AER and other faults, let's add it to the kernel side rather than relying on userspace virtualization to handle it. Signed-off-by:
Alex Williamson <alex.williamson@redhat.com> Reviewed-by:
Eric Auger <eric.auger@redhat.com> Signed-off-by:
-
Alan Brady authored
[ Upstream commit c53d11f6 ] Currently there is a bug in which the PF driver fails to inform clients of a VF reset which then causes clients to leak resources. The bug exists because we were incorrectly checking the I40E_VF_STATE_PRE_ENABLE bit. When a VF is first init we go through a reset to initialize variables and allocate resources but we don't want to inform clients of this first reset since the client isn't fully enabled yet so we set a state bit signifying we're in a "pre-enabled" client state. During the first reset we should be clearing the bit, allowing all following resets to notify the client of the reset when the bit is not set. This patch fixes the issue by negating the 'test_and_clear_bit' check to accurately reflect the behavior we want. Signed-off-by:
Alan Brady <alan.brady@intel.com> Tested-by:
Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by:
-
Dick Kennedy authored
[ Upstream commit 2299e432 ] Warning messages when NVME_TARGET_FC not defined on ppc builds The lpfc_nvmet_replenish_context() function is only meaningful when NVME target mode enabled. Surround the function body with ifdefs for target mode enablement. Signed-off-by:
Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by:
James Smart <james.smart@broadcom.com> Reported-by:
Stephen Rothwell <sfr@canb.auug.org.au> Reviewed-by:
Johannes Thumshirn <jthumshirn@suse.de> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by:
-
Dick Kennedy authored
[ Upstream commit e8bcf0ae ] Local Reject/Invalid RPI errors seen during discovery. Temporary RPI cleanup was occurring regardless of SLI rev. It's only necessary on SLI-4. Adjust the test for whether cleanup is necessary. Signed-off-by:
-
Dick Kennedy authored
[ Upstream commit 184fc2b9 ] Firmware update fails with: status x17 add_status x56 on the final write If multiple DMA buffers are used for the download, some firmware revs have difficulty with signatures and crcs split across the dma buffer boundaries. Resolve by making all writes be a single 4k page in length. Signed-off-by:
-
Jacob Keller authored
[ Upstream commit 3e256ac5 ] We've had support for setting both a minimum and maximum bandwidth via .ndo_set_vf_bw since commit 883a9ccb ("fm10k: Add support for SR-IOV to driver", 2014-09-20). Likely because we do not support minimum rates, the declaration mis-ordered the "unused" parameter, which causes warnings when analyzed with cppcheck. Fix this warning by properly declaring the min_rate and max_rate variables in the declaration and definition (rather than using "unused"). Also rename "rate" to max_rate so as to clarify that we only support setting the maximum rate. Signed-off-by:
-
Nicolas Dechesne authored
[ Upstream commit 46d69e14 ] If the driver is built as a module, autoload won't work because the module alias information is not filled. So user-space can't match the registered device with the corresponding module. Export the module alias information using the MODULE_DEVICE_TABLE() macro. Before this patch: $ modinfo snd_soc_msm8916_analog | grep alias $ After this patch: $ modinfo snd_soc_msm8916_analog | grep alias alias: of:N*T*Cqcom,pm8916-wcd-analog-codecC* alias: of:N*T*Cqcom,pm8916-wcd-analog-codec Signed-off-by:
Nicolas Dechesne <nicolas.dechesne@linaro.org> Signed-off-by:
Mark Brown <broonie@kernel.org> Signed-off-by:
-
Marcelo Ricardo Leitner authored
[ Upstream commit 1ae2eaaa ] As SCTP supports up to 65535 streams, that can lead to very large allocations in sctp_stream_init(). As Xin Long noticed, systems with small amounts of memory are more prone to not have enough memory and dump warnings on dmesg initiated by user actions. Thus, silence them. Also, if the reallocation of stream->out is not necessary, skip it and keep the memory we already have. Reported-by:
Xin Long <lucien.xin@gmail.com> Tested-by:
Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Signed-off-by:
-
Nicholas Piggin authored
[ Upstream commit 80e4d70b ] In xmon, touch_nmi_watchdog() is not expected to be checking that other CPUs have not touched the watchdog, so the code will just call touch_nmi_watchdog() once before re-enabling hard interrupts. Just update our CPU's state, and ignore apparently stuck SMP threads. Arguably touch_nmi_watchdog should check for SMP lockups, and callers should be fixed, but that's not trivial for the input code of xmon. Signed-off-by:
Michael Ellerman <mpe@ellerman.id.au> Signed-off-by:
-
Nicholas Piggin authored
[ Upstream commit 064996d6 ] The SMP hardlockup watchdog cross-checks other CPUs for lockups, which causes xmon headaches because it's assuming interrupts hard disabled means no watchdog troubles. Try to improve that by calling touch_nmi_watchdog() in obvious places where secondaries are spinning. Also annotate these spin loops with spin_begin/end calls. Signed-off-by:
-
Ed Blake authored
[ Upstream commit c7045889 ] Add pm_runtime_get_sync and pm_runtime_put calls to set_fmt callback function. This fixes a bus error during boot when CONFIG_SUSPEND is defined when this function gets called while the device is runtime disabled and device registers are accessed while the clock is disabled. Signed-off-by:
Ed Blake <ed.blake@sondrel.com> Signed-off-by:
-
Jean-François Têtu authored
[ Upstream commit 664611e7 ] The macro used to set the microphone bias level causes the snd_soc_write() call to overwrite other fields in the CDC_A_MICB_1_VAL register. The macro also does not return the proper level value to use. This fixes this by preserving all bits from the register that are not the level while setting the level. Signed-off-by:
Jean-François Têtu <jean-francois.tetu@savoirfairelinux.com> Acked-by:
Srinivas Kandagatla <srinivas.kandagatla@linaro.org> Signed-off-by:
-
Tom Zanussi authored
[ Upstream commit a15f7fc2 ] There are a small number of 'generic fields' (comm/COMM/cpu/CPU) that are found by trace_find_event_field() but are only meant for filtering. Specifically, they unlike normal fields, they have a size of 0 and thus wreak havoc when used as a histogram key. Exclude these (return -EINVAL) when used as histogram keys. Link: http://lkml.kernel.org/r/956154cbc3e8a4f0633d619b886c97f0f0edf7b4.1506105045.git.tom.zanussi@linux.intel.comSigned-off-by:
Tom Zanussi <tom.zanussi@linux.intel.com> Signed-off-by:
Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by:
-
Gabriele Paoloni authored
[ Upstream commit 86acc790 ] Previously, if an non-fatal error was reported by an endpoint, we called report_error_detected() for the endpoint, every sibling on the bus, and their descendents. If any of them did not implement the .error_detected() method, do_recovery() failed, leaving all these devices unrecovered. For example, the system described in the bugzilla below has two devices: 0000:74:02.0 [19e5:a230] SAS controller, driver has .error_detected() 0000:74:03.0 [19e5:a235] SATA controller, driver lacks .error_detected() When a device such as 74:02.0 reported a non-fatal error, do_recovery() failed because 74:03.0 lacked an .error_detected() method. But per PCIe r3.1, sec 6.2.2.2.2, such an error does not compromise the Link and does not affect 74:03.0: Non-fatal errors are uncorrectable errors which cause a particular transaction to be unreliable but the Link is otherwise fully functional. Isolating Non-fatal from Fatal errors provides Requester/Receiver logic in a device or system management software the opportunity to recover from the error without resetting the components on the Link and disturbing other transactions in progress. Devices not associated with the transaction in error are not impacted by the error. Report non-fatal errors only to the endpoint that reported them. We really want to check for AER_NONFATAL here, but the current code structure doesn't allow that. Looking for pci_channel_io_normal is the best we can do now. Link: https://bugzilla.kernel.org/show_bug.cgi?id=197055 Fixes: 6c2b374d ("PCI-Express AER implemetation: AER core and aerdriver") Signed-off-by:
Gabriele Paoloni <gabriele.paoloni@huawei.com> Signed-off-by:
Dongdong Liu <liudongdong3@huawei.com> [bhelgaas: changelog] Signed-off-by:
Bjorn Helgaas <bhelgaas@google.com> Signed-off-by:
-
Jacob Keller authored
[ Upstream commit be664cbe ] Currently, when setting up the IRQ for a q_vector, we set an affinity hint based on the v_idx of that q_vector. Meaning a loop iterates on v_idx, which is an incremental value, and the cpumask is created based on this value. This is a problem in systems with multiple logical CPUs per core (like in simultaneous multithreading (SMT) scenarios). If we disable some logical CPUs, by turning SMT off for example, we will end up with a sparse cpu_online_mask, i.e., only the first CPU in a core is online, and incremental filling in q_vector cpumask might lead to multiple offline CPUs being assigned to q_vectors. Example: if we have a system with 8 cores each one containing 8 logical CPUs (SMT == 8 in this case), we have 64 CPUs in total. But if SMT is disabled, only the 1st CPU in each core remains online, so the cpu_online_mask in this case would have only 8 bits set, in a sparse way. In general case, when SMT is off the cpu_online_mask has only C bits set: 0, 1*N, 2*N, ..., C*(N-1) where C == # of cores; N == # of logical CPUs per core. In our example, only bits 0, 8, 16, 24, 32, 40, 48, 56 would be set. Instead, we should only assign hints for CPUs which are online. Even better, the kernel already provides a function, cpumask_local_spread() which takes an index and returns a CPU, spreading the interrupts across local NUMA nodes first, and then remote ones if necessary. Since we generally have a 1:1 mapping between vectors and CPUs, there is no real advantage to spreading vectors to local CPUs first. In order to avoid mismatch of the default XPS hints, we'll pass -1 so that it spreads across all CPUs without regard to the node locality. Note that we don't need to change the q_vector->affinity_mask as this is initialized to cpu_possible_mask, until an actual affinity is set and then notified back to us. Signed-off-by:
-
Hans de Goede authored
[ Upstream commit 227630cc ] This commit fixes 2 issues with host-wake irq trigger type handling in hci_bcm: 1) bcm_setup_sleep sets sleep_params.host_wake_active based on bcm_device.irq_polarity, but bcm_request_irq was always requesting IRQF_TRIGGER_RISING as trigger type independent of irq_polarity. This was a problem when the irq is described as a GpioInt rather then an Interrupt in the DSDT as for GpioInt-s the value passed to request_irq is honored. This commit fixes this by requesting the correct trigger type depending on bcm_device.irq_polarity. 2) bcm_device.irq_polarity was used to directly store an ACPI polarity value (ACPI_ACTIVE_*). This is undesirable because hci_bcm is also used with device-tree and checking for something like ACPI_ACTIVE_LOW in a non ACPI specific function like bcm_request_irq feels wrong. This commit fixes this by renaming irq_polarity to irq_active_low and changing its type to a bool. Signed-off-by:
Hans de Goede <hdegoede@redhat.com> Signed-off-by:
Marcel Holtmann <marcel@holtmann.org> Signed-off-by:
-
