- 24 Jul, 2014 5 commits
-
-
Gregory CLEMENT authored
Use the common function mvebu_setup_boot_addr_wa() introduced in the commit "ARM: mvebu: Add a common function for the boot address work around" instead of the dedicated version for Armada 375. This commit also moves the workaround in the system-controller module. Indeed the workaround on 375 is really related to setting the boot address which is done by the system controller. As a bonus we no longer use an harcoded value to access the register storing the boot address. Signed-off-by:
Gregory CLEMENT <gregory.clement@free-electrons.com> Signed-off-by:
Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Link: https://lkml.kernel.org/r/1406120453-29291-5-git-send-email-thomas.petazzoni@free-electrons.comSigned-off-by:
Jason Cooper <jason@lakedaemon.net>
-
Gregory CLEMENT authored
On some of the mvebu SoCs and due to internal BootROM issue, the CPU initial jump code must be placed in the SRAM memory of the SoC. In order to achieve this, we have to unmap the BootROM and at some specific location where the BootROM was placed, create a dedicated MBus window for the SRAM. This SRAM is initialized with a few instructions of code that allows to jump to the real secondary CPU boot address. The SRAM used is the Crypto engine one. This work around is currently needed for booting SMP on Armada 375 Z1 and will be needed for cpuidle support on Armada 370. Instead of duplicating the same code, this commit introduces a common function to handle it: mvebu_setup_boot_addr_wa(). Signed-off-by:
-
Gregory CLEMENT authored
Sorting the headers in alphabetic order will help to reduce conflicts when adding new headers later. Signed-off-by:
-
Gregory CLEMENT authored
do_armada_370_xp_cpu_suspend() and armada_370_xp_pmsu_idle_prepare(), have been merged into a single function called armada_370_xp_pmsu_idle_enter() by the commit "bbb92284 ARM: mvebu: slightly refactor/rename PMSU idle related functions", in prepare for the introduction of the CPU hotplug support for Armada XP. But for cpuidle the prepare function will be common to all the mvebu SoCs that use the PMSU, while the suspend function will be specific to each SoC. Keeping the prepare function separate will help reducing code duplication while new SoC support is added. Signed-off-by:
-
Jason Cooper authored
-
- 13 Jul, 2014 1 commit
-
-
Andrew Lunn authored
Now that all boards have been converted to DT and all the support code lives in mach-mvebu, we can remove mach-kirkwood. Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Link: https://lkml.kernel.org/r/1405028192-9623-2-git-send-email-andrew@lunn.chSigned-off-by:
-
- 08 Jul, 2014 2 commits
-
-
Ezequiel Garcia authored
Currently, the coherency fabric support registers two bus notifiers; one for platform, one for pci bus types, with the same notifier block. However, this is illegal and can cause serious issues: the notifier block is also a link in the notifier list and cannot be inserted twice. This commit fixes this by using different notifier blocks (with the same notifier callback) to set the platform and pci bus types notifiers. Fixes: b0063aad ("ARM: mvebu: use hardware I/O coherency also for PCI devices") Reported-by:
Paolo Pisati <p.pisati@gmail.com> Signed-off-by:
Ezequiel Garcia <ezequiel.garcia@free-electrons.com> Link: https://lkml.kernel.org/r/1404826657-6977-1-git-send-email-ezequiel.garcia@free-electrons.comSigned-off-by:
-
Gregory CLEMENT authored
In the inline asm part of the function armada_370_xp_pmsu_idle_enter() the input operand was used. The intent here was to let the compiler choose this register so it could do the optimization it needed. However an input operand is not supposed to be modified by the inline asm code. This can lead to improper generated instructions. In some case generated instruction the compiler made the choice to reuse the same register to store the return value. But in the assembly part this register was modified, so it can lead to return an wrong value. The fix is to use a clobber. Thanks to this the compiler will know that the value of this register will be modified. Signed-off-by:
-
- 01 Jul, 2014 1 commit
-
-
Thomas Petazzoni authored
The SMP boot on Armada 38x and Armada 375 Z1 is currently broken in big-endian configurations, and this commit fixes it for both platforms. For Armada 375 Z1, the problem was in the armada_375_smp_cpu1_enable_code part of the code that gets copied to the Crypto SRAM as a work-around for an issue of the Z1 stepping. This piece of code was not switching the CPU core to big-endian, and not endian-swapping the value read from the Resume Address register (the value is stored little-endian). Due to the introduction of the conditional 'rev r1, r1' instruction, the offset between the 'ldr r0, [pc, #4]' instruction and the value it was looking is different between LE and BE configurations. To solve this, we instead use one 'adr' instruction followed by one 'ldr'. For Armada 38x, the problem was simply that the CPU core was not switched to big endian in the secondary CPU startup function. This change was tested in LE and BE configurations on Armada 385, Armada 375 Z1 and Armada 375 A0. Signed-off-by:
-
- 30 Jun, 2014 8 commits
-
-
Thomas Petazzoni authored
On Marvell Armada XP, when a CPU comes back from deep idle state of cpuidle, it restarts its execution at armada_370_xp_cpu_resume(), which puts back the CPU into the coherency, and then calls the generic cpu_resume() function. While this works on little-endian configurations, it doesn't work on big-endian configurations because the CPU restarts in little-endian, and therefore must be switched back to big-endian to operate properly. To achieve this, a 'setend be' instruction must be executed in big-endian configurations. However, the ARM_BE8() macro that is used to implement nice compile-time conditional for ARM LE vs. ARM BE8 is not easily usable in inline assembly. Therefore, this patch moves the armada_370_xp_cpu_resume() C function, which was anyway just a block of inline assembly, into a proper pmsu_ll.S file, and adds the appropriate ARM_BE8(setend be) instruction. Without this patch, an Armada XP big endian configuration with cpuidle enabled fails to boot, as it hangs as soon as one of the CPU hits the deep idle state. Signed-off-by:
-
Sachin Kamat authored
'mvebu_cpu_reset_init' is local to this file. Signed-off-by:
Sachin Kamat <sachin.kamat@linaro.org> Link: https://lkml.kernel.org/r/1403610235-22654-4-git-send-email-sachin.kamat@samsung.comSigned-off-by:
Sachin Kamat <sachin.kamat@samsung.com> Acked-by:
-
Sachin Kamat authored
'armada_370_xp_cpu_pm_init' is local to this file. Signed-off-by:
-
Sachin Kamat authored
'armada_375_smp_cpu1_enable_wa' is local to this file. Signed-off-by:
-
Gregory CLEMENT authored
On Armada 38x it is possible to get the SoC Id and the revision without using the PCI register. Accessing the PCI registers implies enabling its clock and, because of the initialization issue, not keeping them enable. So if possible it is better to avoid it. Armada 370 and Armada XP provides the SoC ID values from the system controller but not the revision. Armada 375 provides both but the SoC ID value looks buggy (0x6660 instead of 0x6720). Signed-off-by:
-
Gregory CLEMENT authored
Instead of using -1 as error value, use a standard errno. Signed-off-by:
-
Thomas Petazzoni authored
Commit 497a9230 ("ARM: mvebu: implement L2/PCIe deadlock workaround") introduced some logic in coherency.c to adjust the PL310 cache controller Device Tree node of Armada 375 and Armada 38x platform to include the 'arm,io-coherent' property if the system is running with hardware I/O coherency enabled. However, with the L2CC driver cleanup done by Russell King, the initialization of the L2CC driver has been moved earlier, and is now part of the init_IRQ() ARM function in arch/arm/kernel/irq.c. Therefore, calling coherency_init() in ->init_time() is now too late, as the Device Tree property gets added too late (after the L2CC driver has been initialized). In order to fix this, this commit removes the ->init_time() callback use in board-v7.c and replaces it with an ->init_irq() callback. We therefore no longer use the default ->init_irq() callback, but we now use the default ->init_time() callback. In this newly introduced ->init_irq() callback, we call irqchip_init() which is the default behavior when ->init_irq() isn't defined, and then do the initialization related to the coherency: SCU, coherency fabric, and mvebu-mbus (which is needed to start secondary CPUs). Signed-off-by:
-
Thomas Petazzoni authored
In preparation to a small re-organization of the initialization sequence in board-v7.c, this commit moves the registration of the custom external abort handler on Armada 375 later in the boot sequence, and makes it more similar to the other quirks that we already have. There is indeed no need to register this abort handler particularly early, it simply needs to be registered before switching to userspace. In addition to this, this commit makes the registration of the custom abort handler conditional on Armada 375 Z1, because Armada 375 A0 and later iterations are not affected by the issue. This commit was tested on both Armada 375 Z1 and Armada 375 A0 platforms. Signed-off-by:
-
- 24 Jun, 2014 1 commit
-
-
Gregory CLEMENT authored
Wildcards in compatible strings should be avoid. "marvell,armada38x" was recently introduced but was not yet used. The armada 385 SoC is a superset of the armada 380 SoC (with more CPUs and more PCIe slots). So this patch replaces the use of "marvell,armada38x" by the "marvell,armada380" string. Signed-off-by:
-
- 23 Jun, 2014 1 commit
-
-
Sebastian Hesselbarth authored
Commit eeb84545 ("ARM: dts: kirkwood: set Guruplug phy-connection-type to rgmii-id") added phy-connection-type properties to ethernet PHY nodes. Actually, the property has to be set for the ethernet port node instead. Fix it by moving the corresponding properties to the correct nodes. Signed-off-by:
Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com> Link: https://lkml.kernel.org/r/1403555115-13111-1-git-send-email-sebastian.hesselbarth@gmail.com Fixes: eeb84545: ('ARM: dts: kirkwood: set Guruplug phy-connection-type to rgmii-id') Cc: <stable@vger.kernel.org> # v3.16+ Signed-off-by:
-
- 21 Jun, 2014 6 commits
-
-
Ezequiel Garcia authored
Currently, the thermal quirk is skipped only if the SoC revision is known to be one that does not need them, but if the SoC revision cannot be obtained, the quirk is applied assuming it's needed. However, this quirk must be applied only we are sure the SoC needs it, for it breaks the thermal support if applied on a SoC that doesn't need it. The reason for this is that the quirk consists in changing the thermal devicetree compatible string and register offsets, to workaround a hardware bug in the early SoC revision. Such changes are wrong if the SoC is a new revision and doesn't need the workaround. Therefore, this commit changes the behavior, by requiring the SoC revision to be known in order to peform a quirk. Signed-off-by:
-
Thomas Petazzoni authored
Marvell has very recently released a public version of the "Functional specifications" and "Hardware specifications" datasheets for the Marvell Armada 370 SoC. This allows contributors and developers not under NDA with Marvell to get more details about this SoC than what the current kernel code shows, and hopefully allows to improve the support for this SoC in the mainline kernel, as well as in other projects. Signed-off-by:
-
Thomas Petazzoni authored
This commit implements CPU hotplug support for the Marvell Armada XP platform. The CPU hotplug stub functions from hotplug.c are moved into platsmp.c, as it doesn't make much sense to have a separate file just for these two functions. In addition, this commit: * Implements the ->cpu_die() function of SMP operations by calling armada_370_xp_pmsu_idle_enter() to enter the deep idle state for CPUs going offline. * Implements a dummy ->cpu_kill() function, simply needed for the kernel to know we have CPU hotplug support. * The armada_xp_boot_secondary() function makes sure to wake up the CPU if waiting in deep idle state by sending an IPI. This is because armada_xp_boot_secondary() is now used in two different situations: for the initial boot of secondary CPUs (where CPU reset deassert is used to wake up CPUs) and for CPU hotplug (where an IPI is used to take CPU out of deep idle). * At boot time, we exit from the idle state in the ->smp_secondary_init() hook. This commit has been tested using CPU hotplug through sysfs (/sys/devices/system/cpu/cpuX/online) and using kexec. Signed-off-by:
-
Thomas Petazzoni authored
The PMSU idle enter/exit functions will be needed for the CPU hotplug implementation on Armada XP, so this commit removes their static qualifier, and adds the appropriate prototypes in armada-370-xp.h. Signed-off-by:
-
Thomas Petazzoni authored
The CPU hotplug code will need to call into PMSU functions to enter and exit from deep idle states. However, the deep idle state is currently entered by a function called do_armada_370_xp_cpu_suspend() whose name really suggests it's an internal function, but we need to export it to other files in mach-mvebu. Therefore, this commit: * Merges the code of do_armada_370_xp_cpu_suspend() into armada_370_xp_pmsu_idle_prepare(), into a single function called armada_370_xp_pmsu_idle_enter(), which prepares the PMSU for deep idle, and then enters the deep idle state. This code will be common to both cpuidle and CPU hotplug. * For symetry, it renames the armada_370_xp_pmsu_idle_restore() function to armada_370_xp_pmsu_idle_exit(). We also remove the 'noinline' qualifier for these functions, which apparently had no reason to be here. Signed-off-by:
-
Thomas Petazzoni authored
In preparation to the addition of CPU hotplug support for Armada XP, and therefore moving the existing stub functions for hotplug support, this commit removes the reference from the SMP implementation of Armada 375/38x to the armada_xp_cpu_die() function. Proper CPU hotplug support for Armada 375 and 38x will be implemented at a later point. Signed-off-by:
-
- 20 Jun, 2014 4 commits
-
-
Andrew Lunn authored
There is currently no DT binding for the CPLD which controls the LEDs on the Net 2Big and Net 5Big. So use a platform device. Signed-off-by:
Simon Guinot <simon.guinot@sequanux.org> Signed-off-by:
-
Thomas Petazzoni authored
The armada_370_xp_pmsu_idle_prepare() function is only used internally to pmsu.c, so there's no reason to not use the static qualifier. Signed-off-by:
-
Thomas Petazzoni authored
On Marvell Armada platforms, the PMSU (Power Management Service Unit) controls a number of power management related activities, needed for things like suspend/resume, CPU hotplug, cpuidle or even simply SMP. Since cpuidle support was added for Armada XP, the pmsu.c file in arch/arm/mach-mvebu/ calls the cpu_suspend() and cpu_resume() ARM functions, which are only available when CONFIG_ARM_CPU_SUSPEND=y. Therefore, configurations that have CONFIG_ARM_CPU_SUSPEND disabled due to PM_SLEEP being disabled no longer build properly, due to undefined references to cpu_suspend() and cpu_resume(). To fix this, this patch simply ensures CONFIG_ARM_CPU_SUSPEND is always enabled for Marvell EBU v7 platforms. Doing things in a more fine-grained way would require a lot of #ifdef-ery in pmsu.c to isolate the parts that use cpu_suspend()/cpu_resume(), and those parts would anyway have been needed as soon as either one of suspend/resume, CPU hotplug or cpuidle was enabled. Reported-by:
-
Ezequiel Garcia authored
Currently the mvebu boards need to detect the SoC revision in order to apply some quirks needed to workaround issues found on I2C and thermal controllers present only in very early SoC. This detection requires PCI address translation to work, so we need to explicitly select OF_ADDRESS_PCI. This can be considered a partial revert of the following commit, that wrongly removed the option selection: commit 55400f3a Author: Rob Herring <robh@kernel.org> Date: Tue Apr 22 14:15:52 2014 -0500 ARM: mvebu: clean-up unneeded kconfig selects Signed-off-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
- 16 Jun, 2014 4 commits
-
-
Linus Torvalds authored
-
git://git.kernel.org/pub/scm/linux/kernel/git/davem/netLinus Torvalds authored
Pull networking fixes from David Miller: 1) Fix checksumming regressions, from Tom Herbert. 2) Undo unintentional permissions changes for SCTP rto_alpha and rto_beta sysfs knobs, from Denial Borkmann. 3) VXLAN, like other IP tunnels, should advertize it's encapsulation size using dev->needed_headroom instead of dev->hard_header_len. From Cong Wang. * git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: net: sctp: fix permissions for rto_alpha and rto_beta knobs vxlan: Checksum fixes net: add skb_pop_rcv_encapsulation udp: call __skb_checksum_complete when doing full checksum net: Fix save software checksum complete net: Fix GSO constants to match NETIF flags udp: ipv4: do not waste time in __udp4_lib_mcast_demux_lookup vxlan: use dev->needed_headroom instead of dev->hard_header_len MAINTAINERS: update cxgb4 maintainer -
git://git.linaro.org/people/mike.turquette/linuxLinus Torvalds authored
Pull more clock framework updates from Mike Turquette: "This contains the second half the of the clk changes for 3.16. They are simply fixes and code refactoring for the OMAP clock drivers. The sunxi clock driver changes include splitting out the one mega-driver into several smaller pieces and adding support for the A31 SoC clocks" * tag 'clk-for-linus-3.16-part2' of git://git.linaro.org/people/mike.turquette/linux: (25 commits) clk: sunxi: document PRCM clock compatible strings clk: sunxi: add PRCM (Power/Reset/Clock Management) clks support clk: sun6i: Protect SDRAM gating bit clk: sun6i: Protect CPU clock clk: sunxi: Rework clock protection code clk: sunxi: Move the GMAC clock to a file of its own clk: sunxi: Move the 24M oscillator to a file of its own clk: sunxi: Remove calls to clk_put clk: sunxi: document new A31 USB clock compatible clk: sunxi: Implement A31 USB clock ARM: dts: OMAP5/DRA7: use omap5-mpu-dpll-clock capable of dealing with higher frequencies CLK: TI: dpll: support OMAP5 MPU DPLL that need special handling for higher frequencies ARM: OMAP5+: dpll: support Duty Cycle Correction(DCC) CLK: TI: clk-54xx: Set the rate for dpll_abe_m2x2_ck CLK: TI: Driver for DRA7 ATL (Audio Tracking Logic) dt:/bindings: DRA7 ATL (Audio Tracking Logic) clock bindings ARM: dts: dra7xx-clocks: Correct name for atl clkin3 clock CLK: TI: gate: add composite interface clock to OMAP2 only build ARM: OMAP2: clock: add DT boot support for cpufreq_ck CLK: TI: OMAP2: add clock init support ...
-
git://git.infradead.org/users/willy/linux-nvmeLinus Torvalds authored
Pull NVMe update from Matthew Wilcox: "Mostly bugfixes again for the NVMe driver. I'd like to call out the exported tracepoint in the block layer; I believe Keith has cleared this with Jens. We've had a few reports from people who're really pounding on NVMe devices at scale, hence the timeout changes (and new module parameters), hotplug cpu deadlock, tracepoints, and minor performance tweaks" [ Jens hadn't seen that tracepoint thing, but is ok with it - it will end up going away when mq conversion happens ] * git://git.infradead.org/users/willy/linux-nvme: (22 commits) NVMe: Fix START_STOP_UNIT Scsi->NVMe translation. NVMe: Use Log Page constants in SCSI emulation NVMe: Define Log Page constants NVMe: Fix hot cpu notification dead lock NVMe: Rename io_timeout to nvme_io_timeout NVMe: Use last bytes of f/w rev SCSI Inquiry NVMe: Adhere to request queue block accounting enable/disable NVMe: Fix nvme get/put queue semantics NVMe: Delete NVME_GET_FEAT_TEMP_THRESH NVMe: Make admin timeout a module parameter NVMe: Make iod bio timeout a parameter NVMe: Prevent possible NULL pointer dereference NVMe: Fix the buffer size passed in GetLogPage(CDW10.NUMD) NVMe: Update data structures for NVMe 1.2 NVMe: Enable BUILD_BUG_ON checks NVMe: Update namespace and controller identify structures to the 1.1a spec NVMe: Flush with data support NVMe: Configure support for block flush NVMe: Add tracepoints NVMe: Protect against badly formatted CQEs ...
-
- 15 Jun, 2014 7 commits
-
-
Daniel Borkmann authored
Commit 3fd091e7 ("[SCTP]: Remove multiple levels of msecs to jiffies conversions.") has silently changed permissions for rto_alpha and rto_beta knobs from 0644 to 0444. The purpose of this was to discourage users from tweaking rto_alpha and rto_beta knobs in production environments since they are key to correctly compute rtt/srtt. RFC4960 under section 6.3.1. RTO Calculation says regarding rto_alpha and rto_beta under rule C3 and C4: [...] C3) When a new RTT measurement R' is made, set RTTVAR <- (1 - RTO.Beta) * RTTVAR + RTO.Beta * |SRTT - R'| and SRTT <- (1 - RTO.Alpha) * SRTT + RTO.Alpha * R' Note: The value of SRTT used in the update to RTTVAR is its value before updating SRTT itself using the second assignment. After the computation, update RTO <- SRTT + 4 * RTTVAR. C4) When data is in flight and when allowed by rule C5 below, a new RTT measurement MUST be made each round trip. Furthermore, new RTT measurements SHOULD be made no more than once per round trip for a given destination transport address. There are two reasons for this recommendation: First, it appears that measuring more frequently often does not in practice yield any significant benefit [ALLMAN99]; second, if measurements are made more often, then the values of RTO.Alpha and RTO.Beta in rule C3 above should be adjusted so that SRTT and RTTVAR still adjust to changes at roughly the same rate (in terms of how many round trips it takes them to reflect new values) as they would if making only one measurement per round-trip and using RTO.Alpha and RTO.Beta as given in rule C3. However, the exact nature of these adjustments remains a research issue. [...] While it is discouraged to adjust rto_alpha and rto_beta and not further specified how to adjust them, the RFC also doesn't explicitly forbid it, but rather gives a RECOMMENDED default value (rto_alpha=3, rto_beta=2). We have a couple of users relying on the old permissions before they got changed. That said, if someone really has the urge to adjust them, we could allow it with a warning in the log. Fixes: 3fd091e7 ("[SCTP]: Remove multiple levels of msecs to jiffies conversions.") Signed-off-by:
Daniel Borkmann <dborkman@redhat.com> Cc: Vlad Yasevich <vyasevich@gmail.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
David S. Miller authored
Tom Herbert says: ==================== Fixes related to some recent checksum modifications. - Fix GSO constants to match NETIF flags - Fix logic in saving checksum complete in __skb_checksum_complete - Call __skb_checksum_complete from UDP if we are checksumming over whole packet in order to save checksum. - Fixes to VXLAN to work correctly with checksum complete ==================== Signed-off-by: -
Tom Herbert authored
Call skb_pop_rcv_encapsulation and postpull_rcsum for the Ethernet header to work properly with checksum complete. Signed-off-by:
Tom Herbert <therbert@google.com> Signed-off-by:
-
Tom Herbert authored
This function is used by UDP encapsulation protocols in RX when crossing encapsulation boundary. If ip_summed is set to CHECKSUM_UNNECESSARY and encapsulation is not set, change to CHECKSUM_NONE since the checksum has not been validated within the encapsulation. Clears csum_valid by the same rationale. Signed-off-by:
-
Tom Herbert authored
In __udp_lib_checksum_complete check if checksum is being done over all the data (len is equal to skb->len) and if it is call __skb_checksum_complete instead of __skb_checksum_complete_head. This allows checksum to be saved in checksum complete. Signed-off-by:
-
Tom Herbert authored
Geert reported issues regarding checksum complete and UDP. The logic introduced in commit 7e3cead5 ("net: Save software checksum complete") is not correct. This patch: 1) Restores code in __skb_checksum_complete_header except for setting CHECKSUM_UNNECESSARY. This function may be calculating checksum on something less than skb->len. 2) Adds saving checksum to __skb_checksum_complete. The full packet checksum 0..skb->len is calculated without adding in pseudo header. This value is saved in skb->csum and then the pseudo header is added to that to derive the checksum for validation. 3) In both __skb_checksum_complete_header and __skb_checksum_complete, set skb->csum_valid to whether checksum of zero was computed. This allows skb_csum_unnecessary to return true without changing to CHECKSUM_UNNECESSARY which was done previously. 4) Copy new csum related bits in __copy_skb_header. Reported-by:
Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by:
-
Tom Herbert authored
Joseph Gasparakis reported that VXLAN GSO offload stopped working with i40e device after recent UDP changes. The problem is that the SKB_GSO_* bits are out of sync with the corresponding NETIF flags. This patch fixes that. Also, we add BUILD_BUG_ONs in net_gso_ok for several GSO constants that were missing to avoid the problem in the future. Reported-by:
Joseph Gasparakis <joseph.gasparakis@intel.com> Signed-off-by:
-
