aboutsummaryrefslogtreecommitdiff
path: root/drivers
AgeCommit message (Collapse)AuthorFilesLines
2017-09-27bcache: use llist_for_each_entry_safe() in __closure_wake_up()Coly Li1-2/+2
Commit 09b3efec ("bcache: Don't reinvent the wheel but use existing llist API") replaces the following while loop by llist_for_each_entry(), - - while (reverse) { - cl = container_of(reverse, struct closure, list); - reverse = llist_next(reverse); - + llist_for_each_entry(cl, reverse, list) { closure_set_waiting(cl, 0); closure_sub(cl, CLOSURE_WAITING + 1); } This modification introduces a potential race by iterating a corrupted list. Here is how it happens. In the above modification, closure_sub() may wake up a process which is waiting on reverse list. If this process decides to wait again by calling closure_wait(), its cl->list will be added to another wait list. Then when llist_for_each_entry() continues to iterate next node, it will travel on another new wait list which is added in closure_wait(), not the original reverse list in __closure_wake_up(). It is more probably to happen on UP machine because the waked up process may preempt the process which wakes up it. Use llist_for_each_entry_safe() will fix the issue, the safe version fetch next node before waking up a process. Then the copy of next node will make sure list iteration stays on original reverse list. Fixes: 09b3efec81de ("bcache: Don't reinvent the wheel but use existing llist API") Signed-off-by: Coly Li <[email protected]> Reported-by: Michael Lyle <[email protected]> Reviewed-by: Byungchul Park <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
2017-09-28Merge branch 'drm-fixes-4.14' of git://people.freedesktop.org/~agd5f/linux ↵Dave Airlie2-2/+189
into drm-fixes A few fixes for 4.14. Nothing too major. * 'drm-fixes-4.14' of git://people.freedesktop.org/~agd5f/linux: drm/radeon: disable hard reset in hibernate for APUs drm/amdgpu: revert tile table update for oland
2017-09-28Merge branch 'etnaviv/fixes' of https://git.pengutronix.de/git/lst/linux ↵Dave Airlie2-2/+7
into drm-fixes Just two small etnaviv fixes, one fixing a list corruption, the other fixing a NULL ptr deref in an error path. * 'etnaviv/fixes' of https://git.pengutronix.de/git/lst/linux: etnaviv: fix gem object list corruption etnaviv: fix submit error path
2017-09-28Merge tag 'drm-amdkfd-fixes-2017-09-24' of ↵Dave Airlie4-5/+23
git://people.freedesktop.org/~gabbayo/linux into drm-fixes It contains the following fixes: - correct checking of return value - send correct parameter to function (According to the parameter type) - avoid spamming of dmesg log - fix queue wrapping calculations * tag 'drm-amdkfd-fixes-2017-09-24' of git://people.freedesktop.org/~gabbayo/linux: drm/amdkfd: Print event limit messages only once per process drm/amdkfd: Fix kernel-queue wrapping bugs drm/amdkfd: Fix incorrect destroy_mqd parameter drm/amdkfd: check for null dev to avoid a null pointer dereference
2017-09-27mtd: nand: atmel: fix buffer overflow in atmel_pmecc_userRichard Genoud1-1/+1
When calculating the size needed by struct atmel_pmecc_user *user, the dmu and delta buffer sizes were forgotten. This lead to a memory corruption (especially with a large ecc_strength). Link: http://lkml.kernel.org/r/[email protected] Fixes: f88fc122cc34 ("mtd: nand: Cleanup/rework the atmel_nand driver") Cc: [email protected] Reported-by: Richard Genoud <[email protected]> Pointed-at-by: Boris Brezillon <[email protected]> Signed-off-by: Richard Genoud <[email protected]> Reviewed-by: Nicolas Ferre <[email protected]> Signed-off-by: Boris Brezillon <[email protected]>
2017-09-27IB/hfi1: Unsuccessful PCIe caps tuning should not fail driver loadHarish Chegondi1-29/+21
Failure to tune PCIe capabilities should not fail driver load. This can cause the driver load to fail on systems with any of the following: 1. HFI's parent is not root. Example: HFI card is behind a PCIe bridge. 2. HFI's parent is not PCI Express capable. In these situations, failure to tune PCIe capabilities should be logged in the system message logs but not cause the driver load to fail. This patch also ensures pcie capability word DevCtl is written only after a successful read and the capability tuning process continues even if read/write of the pcie capability word DevCtl fails. Fixes: c53df62c7a9a ("IB/hfi1: Check return values from PCI config API calls") Fixes: bf70a7757736 ("staging/rdma/hfi1: Enable WFR PCIe extended tags from the driver") Reviewed-by: Michael J. Ruhl <[email protected]> Reviewed-by: Mike Marciniszyn <[email protected]> Reviewed-by: Jakub Byczkowski <[email protected]> Signed-off-by: Harish Chegondi <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27IB/hfi1: On error, fix use after free during user context setupMichael J. Ruhl1-19/+22
During base context setup, if setup_base_ctxt() fails, the context is deallocated. This is incorrect because the context is referenced on return, to notify any waiting subcontext. If there are no subcontexts the pointer will be invalid. Reorganize the error path so that deallocate_ctxt() is called after all the possible subcontexts have been notified. Reviewed-by: Ira Weiny <[email protected]> Signed-off-by: Michael J. Ruhl <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27Revert "IB/ipoib: Update broadcast object if PKey value was changed in index 0"Alex Estrin1-13/+0
commit 9a9b8112699d will cause core to fail UD QP from being destroyed on ipoib unload, therefore cause resources leakage. On pkey change event above patch modifies mgid before calling underlying driver to detach it from QP. Drivers' detach_mcast() will fail to find modified mgid it was never given to attach in a first place. Core qp->usecnt will never go down, so ib_destroy_qp() will fail. IPoIB driver actually does take care of new broadcast mgid based on new pkey by destroying an old mcast object in ipoib_mcast_dev_flush()) .... if (priv->broadcast) { rb_erase(&priv->broadcast->rb_node, &priv->multicast_tree); list_add_tail(&priv->broadcast->list, &remove_list); priv->broadcast = NULL; } ... then in restarted ipoib_macst_join_task() creating a new broadcast mcast object, sending join request and on completion tells the driver to attach to reinitialized QP: ... if (!priv->broadcast) { ... broadcast = ipoib_mcast_alloc(dev, 0); ... memcpy(broadcast->mcmember.mgid.raw, priv->dev->broadcast + 4, sizeof (union ib_gid)); priv->broadcast = broadcast; ... Fixes: 9a9b8112699d ("IB/ipoib: Update broadcast object if PKey value was changed in index 0") Cc: [email protected] Reviewed-by: Mike Marciniszyn <[email protected]> Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Alex Estrin <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Reviewed-by: Feras Daoud <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27IB/hfi1: Return correct value in general interrupt handlerKamenee Arumugam1-1/+3
The general interrupt handler returns IRQ_HANDLED whether an IRQ was handled or not. Determine if an IRQ was handled and return the correct value. Reviewed-by: Dennis Dalessandro <[email protected]> Reviewed-by: Michael J. Ruhl <[email protected]> Signed-off-by: Kamenee Arumugam <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27IB/hfi1: Check eeprom config partition validityJan Sokolowski1-5/+15
Relying on a trailing magic value is incorrect. There are instances where this is not present as trailing magic value has a specific purpose which is not partition validation. Instead use the header magic value which is present in all variants of the platform configuration and is intended for validation. This is also used in other locations in the driver. Fixes: bc5214ee2922 (IB/hfi1: Handle missing magic values in config file) Reviewed-by: Jakub Byczkowski <[email protected]> Signed-off-by: Jan Sokolowski <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27IB/hfi1: Only reset QSFP after link up and turn off AOC TXSebastian Sanchez3-3/+15
QSFP reset enables AOC transmitters by default. They should be off before moving to high power mode to complete the setup. There is no need to reset the QSFP during LNI failure as it was reset at link down. Reviewed-by: Mike Marciniszyn <[email protected]> Reviewed-by: Jakub Byczkowski <[email protected]> Signed-off-by: Sebastian Sanchez <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27IB/hfi1: Turn off AOC TX after offline substatesSebastian Sanchez2-21/+65
Offline.quietDuration was added in the 8051 firmware, and the driver only turns off the AOC transmitters when offline.quiet is reached. However, the AOC transmitters need to be turned off at the new state. Therefore, turn off the AOC transmitters at any offline substates including offline.quiet and offline.quietDuration, then recheck we reached offline.quiet to support backwards compatibility. Reviewed-by: Jakub Byczkowski <[email protected]> Reviewed-by: Mike Marciniszyn <[email protected]> Signed-off-by: Sebastian Sanchez <[email protected]> Signed-off-by: Dennis Dalessandro <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2017-09-27iommu/amd: pr_err() strings should end with newlinesArvind Yadav1-4/+4
pr_err() messages should end with a new-line to avoid other messages being concatenated. So replace '/n' with '\n'. Signed-off-by: Arvind Yadav <[email protected]> Fixes: 45a01c42933b ('iommu/amd: Add function copy_dev_tables()') Signed-off-by: Joerg Roedel <[email protected]>
2017-09-27iommu/mediatek: Limit the physical address in 32bit for v7sYong Wu1-1/+2
The ARM short descriptor has already limited the physical address to 32bit after the commit <76557391433c> ("iommu/io-pgtable: Sanitise map/unmap addresses"). But in MediaTek 4GB mode, the physical address is from 0x1_0000_0000 to 0x1_ffff_ffff. this will cause: WARNING: CPU: 4 PID: 3900 at xxx/drivers/iommu/io-pgtable-arm-v7s.c:482 arm_v7s_map+0x40/0xf8 Modules linked in: CPU: 4 PID: 3900 Comm: weston Tainted: G S W 4.9.44 #1 Hardware name: MediaTek MT2712m1v1 board (DT) task: ffffffc0eaa5b280 task.stack: ffffffc0e9858000 PC is at arm_v7s_map+0x40/0xf8 LR is at mtk_iommu_map+0x64/0x90 pc : [<ffffff80085b09e8>] lr : [<ffffff80085b29fc>] pstate: 000001c5 sp : ffffffc0e985b920 x29: ffffffc0e985b920 x28: 0000000127d00000 x27: 0000000000100000 x26: ffffff8008f9e000 x25: 0000000000000003 x24: 0000000000100000 x23: 0000000127d00000 x22: 00000000ff800000 x21: ffffffc0f7ec8ce0 x20: 0000000000000003 x19: 0000000000000003 x18: 0000000000000002 x17: 0000007f7e5d72c0 x16: ffffff80082b0f08 x15: 0000000000000001 x14: 000000000000003f x13: 0000000000000000 x12: 0000000000000028 x11: 0088000000000000 x10: 0000000000000000 x9 : ffffff80092fa000 x8 : ffffffc0e9858000 x7 : ffffff80085b29d8 x6 : 0000000000000000 x5 : ffffff80085b09a8 x4 : 0000000000000003 x3 : 0000000000100000 x2 : 0000000127d00000 x1 : 00000000ff800000 x0 : 0000000000000001 ... Call trace: [<ffffff80085b09e8>] arm_v7s_map+0x40/0xf8 [<ffffff80085b29fc>] mtk_iommu_map+0x64/0x90 [<ffffff80085ab5f8>] iommu_map+0x100/0x3a0 [<ffffff80085ab99c>] default_iommu_map_sg+0x104/0x168 [<ffffff80085aead8>] iommu_dma_alloc+0x238/0x3f8 [<ffffff8008098b30>] __iommu_alloc_attrs+0xa8/0x260 [<ffffff80085f364c>] mtk_drm_gem_create+0xac/0x180 [<ffffff80085f3894>] mtk_drm_gem_dumb_create+0x54/0xc8 [<ffffff80085d576c>] drm_mode_create_dumb_ioctl+0xa4/0xd8 [<ffffff80085cb2a0>] drm_ioctl+0x1c0/0x490 In order to satify this, Limit the physical address to 32bit. Signed-off-by: Yong Wu <[email protected]> Acked-by: Will Deacon <[email protected]> Signed-off-by: Joerg Roedel <[email protected]>
2017-09-27iommu/io-pgtable-arm-v7s: Need dma-sync while there is no QUIRK_NO_DMAYong Wu1-1/+1
Fix the commit 81b3c2521844 ("iommu/io-pgtable: Introduce explicit coherency"). If there is no IO_PGTABLE_QUIRK_NO_DMA, we should call dma_sync_single_for_device for cache synchronization. Signed-off-by: Yong Wu <[email protected]> Fixes: 81b3c2521844 ('iommu/io-pgtable: Introduce explicit coherency') Reviewed-by: Robin Murphy <[email protected]> Signed-off-by: Joerg Roedel <[email protected]>
2017-09-27pinctrl/amd: Fix build dependency on pinmux codePetr Mladek1-0/+1
The commit 79d2c8bede2c93f943 ("pinctrl/amd: save pin registers over suspend/resume") caused the following compilation errors: drivers/pinctrl/pinctrl-amd.c: In function ‘amd_gpio_should_save’: drivers/pinctrl/pinctrl-amd.c:741:8: error: ‘const struct pin_desc’ has no member named ‘mux_owner’ if (pd->mux_owner || pd->gpio_owner || ^ drivers/pinctrl/pinctrl-amd.c:741:25: error: ‘const struct pin_desc’ has no member named ‘gpio_owner’ if (pd->mux_owner || pd->gpio_owner || We need to enable CONFIG_PINMUX for this driver as well. Cc: [email protected] Fixes: 79d2c8bede2c93f943 ("pinctrl/amd: save pin registers over suspend/resume") Signed-off-by: Petr Mladek <[email protected]> Signed-off-by: Linus Walleij <[email protected]>
2017-09-27mtd: Fix partition alignment check on multi-erasesize devicesBoris Brezillon1-0/+8
Commit 1eeef2d7483a ("mtd: handle partitioning on devices with 0 erasesize") introduced a regression on heterogeneous erase region devices. Alignment of the partition was tested against the master eraseblock size which can be bigger than the slave one, thus leading to some partitions being marked as read-only. Update wr_alignment to match this slave erasesize after this erasesize has been determined by picking the biggest erasesize of all the regions embedded in the MTD partition. Reported-by: Mathias Thore <[email protected]> Fixes: 1eeef2d7483a ("mtd: handle partitioning on devices with 0 erasesize") Cc: <[email protected]> Signed-off-by: Boris Brezillon <[email protected]> Tested-by: Mathias Thore <[email protected]> Reviewed-by: Mathias Thore <[email protected]>
2017-09-27platform/x86: fujitsu-laptop: Don't oops when FUJ02E3 is not presntVille Syrjälä1-4/+6
My Fujitsu-Siemens Lifebook S6120 doesn't have the FUJ02E3 device, but it does have FUJ02B1. That means we do register the backlight device (and it even seems to work), but the code will oops as soon as we try to set the backlight brightness because it's trying to call call_fext_func() with a NULL device. Let's just skip those function calls when the FUJ02E3 device is not present. Cc: Jonathan Woithe <[email protected]> Cc: Andy Shevchenko <[email protected]> Signed-off-by: Ville Syrjälä <[email protected]> Cc: <[email protected]> # 4.13.x Signed-off-by: Darren Hart (VMware) <[email protected]>
2017-09-26Merge tag 'wireless-drivers-for-davem-2017-09-25' of ↵David S. Miller17-46/+125
git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers Kalle Valo says: ==================== wireless-drivers fixes for 4.14 Quite a lot of fixes this time. Most notable is the brcmfmac fix for a CVE issue. iwlwifi * a couple of bugzilla bugs related to multicast handling * two fixes for WoWLAN bugs that were causing queue hangs and re-initialization problems * two fixes for potential uninitialized variable use reported by Dan Carpenter in relation to a recently introduced patch * a fix for buffer reordering in the newly supported 9000 device family * fix a race when starting aggregation * small fix for a recent patch to wake mac80211 queues * send non-bufferable management frames in the generic queue so they are not sent on queues that are under power-save ath10k * fix a PCI PM related gcc warning brcmfmac * CVE-2017-0786: add length check scan results from firmware * respect passive scan requests from user space qtnfmac * fix race in tx path when using multiple interfaces * cancel ongoing scan when removing the wireless interface ==================== Signed-off-by: David S. Miller <[email protected]>
2017-09-26Merge tag 'mmc-v4.14-rc1' of ↵Linus Torvalds2-47/+15
git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc Pull MMC fixes from Ulf Hansson: - sdhci-pci: Fix voltage switch for some Intel host controllers - tmio: remove broken and noisy debug macro * tag 'mmc-v4.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc: mmc: sdhci-pci: Fix voltage switch for some Intel host controllers mmc: tmio: remove broken and noisy debug macro
2017-09-26atlantic: fix iommu errorsPavel Belous3-27/+49
Call skb_frag_dma_map multiple times if tx length is greater than device max and avoid processing tx ring until entire packet has been sent. Signed-off-by: Igor Russkikh <[email protected]> Signed-off-by: Pavel Belous <[email protected]> Signed-off-by: David S. Miller <[email protected]>
2017-09-26aquantia: Fix transient invalid link down/up indicationsIgor Russkikh1-2/+1
Due to a bug in aquantia atlantic card firmware, it sometimes reports invalid link speed bits. That caused driver to report link down events, although link itself is totally fine. This patch ignores such out of blue readings. Signed-off-by: Pavel Belous <[email protected]> Signed-off-by: Igor Russkikh <[email protected]> Signed-off-by: David S. Miller <[email protected]>
2017-09-26aquantia: Fix Tx queue hangupsIgor Russkikh6-59/+76
Driver did a poor job in managing its Tx queues: Sometimes it could stop tx queues due to link down condition in aq_nic_xmit - but never waked up them. That led to Tx path total suspend. This patch fixes this and improves generic queue management: - introduces queue restart counter - uses generic netif_ interface to disable and enable tx path - refactors link up/down condition and introduces dmesg log event when link changes. - introduces new constant for minimum descriptors count required for queue wakeup Signed-off-by: Pavel Belous <[email protected]> Signed-off-by: Igor Russkikh <[email protected]> Signed-off-by: David S. Miller <[email protected]>
2017-09-26aquantia: Setup max_mtu in ndev to enable jumbo framesIgor Russkikh2-10/+3
Although hardware is capable for almost 16K MTU, without max_mtu field correctly set it only allows standard MTU to be used. This patch enables max MTU, calculating it from hardware maximum frame size of 16352 octets (including FCS). Fixes: 5513e16421cb ("net: ethernet: aquantia: Fixes for aq_ndev_change_mtu") Signed-off-by: Pavel Belous <[email protected]> Signed-off-by: Igor Russkikh <[email protected]> Signed-off-by: David S. Miller <[email protected]>
2017-09-26drm/i915/bios: ignore HDMI on port AJani Nikula1-0/+7
The hardware state readout oopses after several warnings when trying to use HDMI on port A, if such a combination is configured in VBT. Filter the combo out already at the VBT parsing phase. v2: also ignore DVI (Ville) Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102889 Cc: [email protected] Cc: Imre Deak <[email protected]> Reviewed-by: Ville Syrjälä <[email protected]> Tested-by: Daniel Drake <[email protected]> Signed-off-by: Jani Nikula <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected] (cherry picked from commit d27ffc1d00327c29b3aa97f941b42f0949f9e99f) Signed-off-by: Rodrigo Vivi <[email protected]>
2017-09-26drm/i915: remove redundant variable hw_checkColin Ian King1-2/+0
hw_check is being assigned and updated but is no longer being read, hence it is redundant and can be removed. Detected by clang scan-build: "warning: Value stored to 'hw_check' during its initialization is never read" Fixes: f6d1973db2d2 ("drm/i915: Move modeset state verifier calls") Signed-off-by: Colin Ian King <[email protected]> Reviewed-by: Chris Wilson <[email protected]> Signed-off-by: Daniel Vetter <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected] (cherry picked from commit 4babc5e27cfda59e2e257d28628b8d853aea5206) Signed-off-by: Rodrigo Vivi <[email protected]>
2017-09-26drm/i915: always update ELD connector type after get modesJani Nikula2-5/+17
drm_edid_to_eld() initializes the connector ELD to zero, overwriting the ELD connector type initialized in intel_audio_codec_enable(). If userspace does getconnector and thus get_modes after modeset, a subsequent audio component i915_audio_component_get_eld() call will receive an ELD without the connector type properly set. It's fine for HDMI, but screws up audio for DP. Always set the ELD connector type at intel_connector_update_modes() based on the connector type. We can drop the connector type update from intel_audio_codec_enable(). Credits to Joseph Nuzman <[email protected]> for figuring this out. Cc: Ville Syrjälä <[email protected]> Cc: Joseph Nuzman <[email protected]> Reported-by: Joseph Nuzman <[email protected]> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101583 Reviewed-by: Ville Syrjälä <[email protected]> Tested-by: Joseph Nuzman <[email protected]> Cc: [email protected] # v4.10+, maybe earlier Signed-off-by: Jani Nikula <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected] (cherry picked from commit d81fb7fd9436e81fda67e5bc8ed0713aa28d3db2) Signed-off-by: Rodrigo Vivi <[email protected]>
2017-09-26phy: rockchip-typec: Don't set the aux voltage swing to 400 mVDouglas Anderson1-3/+4
On rk3399-gru-kevin there are some cases where we're seeing AUX CH failures when trying to do DisplayPort over type C. Problems are intermittent and don't reproduce all the time. Problems are often bursty and failures persist for several seconds before going away. The failure case I focused on is: * A particular type C to HDMI adapter. * One orientation (flip mode) of that adapter. * Easier to see failures when something is plugged into the _other type C port at the same time. * Problems reproduce on both type C ports (left and right side). Ironically problems also stop reproducing when I solder wires onto the AUX CH signals on a port (even if no scope is connected to the signals). In this case, problems only stop reproducing on the port with the wires connected. From the above it appears that something about the signaling on the aux channel is marginal and any slight differences can bring us over the edge to failure. It turns out that we can fix our problems by just increasing the voltage swing of the AUX CH, giving us a bunch of extra margin. In DP up to version 1.2 the voltage swing on the aux channel was specced as .29 V to 1.38 V. In DP version 1.3 the aux channel voltage was tightened to be between .29 V and .40 V, but it clarifies that it really only needs the lower voltage when operating at the highest speed (HBR3 mode). So right now we are trying to use a voltage that technically should be valid for all versions of the spec (including version 1.3 when transmitting at HBR3). That would be great to do if it worked reliably. ...but it doesn't seem to. It turns out that if you continue to read through the DP part of the rk3399 TRM and other parts of the type C PHY spec you'll find out that while the rk3399 does support DP 1.3, it doesn't support HBR3. The docs specifically say "RBR, HBR and HBR2 data rates only". Thus there is actually no requirement to support an AUX CH swing of .4 V. Even if there is no actual requirement to support the tighter voltage swing, one could possibly argue that we should support it anyway. The DP spec clarifies that the lower voltage on the AUX CH will reduce cross talk in some cases and that seems like it could be beneficial even at the lower bit rates. At the moment, though, we are seeing problems with the AUX CH and not on the other lines. Also, checking another known working and similar laptop shows that the other laptop runs the AUX channel at a higher voltage. Other notes: * Looking at measurements done on the AUX CH we weren't actually compliant with the DP 1.3 spec anyway. AUX CH peek-to-peek voltage was measured on rk3399-gru-kevin as .466 V which is > .4 V. * With this new patch the AUX channel isn't actually 1.0 V, but it has been confirmed that the signal is better and has more margin. Eye diagram passes. * If someone were truly an expert in the Type C PHY and in DisplayPort signaling they might be able to make things work and keep the voltage at < .4 V. The Type C PHY seems to have a plethora of tuning knobs that could almost certainly improve the signal integrity. Some of these things (like enabling tx_fcm_full_margin) even seem to fix my problems. However, lacking expertise I can't say whether this is a better or worse solution. Tightening signals to give cleaner waveforms can often have adverse affects, like increasing EMI or adding noise to other signals. I'd rather not tune things like this without a healthy application of expertise that I don't have. Signed-off-by: Douglas Anderson <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26phy: rockchip-typec: Set the AUX channel flip state earlierDouglas Anderson1-20/+42
On some DP monitors we found that setting the wrong flip state on the AUX channel could cause the monitor to stop asserting HotPlug Detect (HPD). Setting the right flip state caused these monitors to start asserting HotPlug Detect again. Here's what we believe was happening: * We'd plug in the monitor and we'd see HPD assert * We'd quickly see HPD deassert * The kernel would try to init the type C PHY but would init it in USB mode (because there was a peripheral there but no HPD) * Because the kernel never set the flip mode properly we'd never see the HPD come back. With this change, we'll still see HPD disappear (we don't think there's anything we can do about that), but then it will come back. Overall we can say that it's sane to set the AUX channel flip state even when HPD is not asserted. NOTE: to make this change possible, I needed to do a bit of cleanup to the tcphy_dp_aux_calibration() function so that it doesn't ever clobber the FLIP state. This made it very obvious that a line of code documented as "setting bit 12" also did a bunch of other magic, undocumented stuff. For now I'll just break out the bits and add a comment that this is black magic and we'll try to document tcphy_dp_aux_calibration() better in a future CL. ALSO NOTE: the old function used to write a bunch of hardcoded values in _some_ cases instead of doing a read-modify-write. One could possibly assert that these could have had (beneficial) side effects and thus with this new code (which always does read-modify-write) we could have a bug. We shouldn't need to worry, though, since in the old code tcphy_dp_aux_calibration() was always called following the de-assertion of "reset" the the type C PHY. ...so the type C PHY was always in default state. TX_ANA_CTRL_REG_1 is documented to be 0x0 after reset. This was also confirmed by printk. Suggested-by: Shawn Nematbakhsh <[email protected]> Reviewed-by: Chris Zhong <[email protected]> Signed-off-by: Douglas Anderson <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26phy: mvebu-cp110: checking for NULL instead of IS_ERR()Dan Carpenter1-2/+2
devm_ioremap_resource() never returns NULL, it only returns error pointers so this test needs to be changed. Fixes: d0438bd6aa09 ("phy: add the mvebu cp110 comphy driver") Signed-off-by: Dan Carpenter <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26phy: mvebu-cp110-comphy: explicitly set the pipe selectorAntoine Tenart1-0/+10
The pipe selector is used to select some modes (such as USB or PCIe). Otherwise it must be set to 0 (or "unconnected"). This patch does this to ensure it is not set to an incompatible value when using the supported modes (SGMII, 10GKR). Signed-off-by: Antoine Tenart <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26phy: mvebu-cp110-comphy: fix mux error checkAntoine Tenart1-2/+2
The mux value is retrieved from the mvebu_comphy_get_mux() function which returns an int. In mvebu_comphy_power_on() this int is stored to a u32 and a check is made to ensure it's not negative. Which is wrong. This fixes it. Fixes: d0438bd6aa09 ("phy: add the mvebu cp110 comphy driver") Signed-off-by: Antoine Tenart <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26phy: phy-mtk-tphy: fix NULL point of chip bankChunfeng Yun1-1/+2
Chip bank of version-1 is initialized as NULL, but it's used by pcie_phy_instance_power_on/off(), so assign it a right address. Signed-off-by: Chunfeng Yun <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26phy: tegra: Handle return value of kasprintfArvind Yadav1-0/+2
kasprintf() can fail and it's return value must be checked. Signed-off-by: Arvind Yadav <[email protected]> Signed-off-by: Kishon Vijay Abraham I <[email protected]>
2017-09-26drm/tegra: trace: Fix path to includeThierry Reding1-1/+1
The TRACE_INCLUDE_FILE macro needs to specify the path relative to the define_trace.h header rather than relative to the file defining it. Reported-by: Dmitry Osipenko <[email protected]> Tested-by: Dmitry Osipenko <[email protected]> Signed-off-by: Thierry Reding <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-09-25rocker: fix rocker_tlv_put_* functions for KASANArnd Bergmann1-18/+30
Inlining these functions creates lots of stack variables that each take 64 bytes when KASAN is enabled, leading to this warning about potential stack overflow: drivers/net/ethernet/rocker/rocker_ofdpa.c: In function 'ofdpa_cmd_flow_tbl_add': drivers/net/ethernet/rocker/rocker_ofdpa.c:621:1: error: the frame size of 2752 bytes is larger than 1536 bytes [-Werror=frame-larger-than=] gcc-8 can now consolidate the stack slots itself, but on older versions we get the same behavior by using a temporary variable that holds a copy of the inline function argument. Cc: [email protected] Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81715 Signed-off-by: Arnd Bergmann <[email protected]> Signed-off-by: David S. Miller <[email protected]>
2017-09-25scsi: scsi_transport_fc: Also check for NOTPRESENT in fc_remote_port_add()Hannes Reinecke1-1/+2
During failover there is a small race window between fc_remote_port_add() and fc_timeout_deleted_rport(); the latter drops the lock after setting the port to NOTPRESENT, so if fc_remote_port_add() is called right at that time it will fail to detect the existing rport and happily adding a new structure, causing rports to get registered twice. Signed-off-by: Hannes Reinecke <[email protected]> Reviewed-by: Johannes Thumshirn <[email protected]> Signed-off-by: Martin K. Petersen <[email protected]>
2017-09-25PCI: Fix race condition with driver_overrideNicolai Stange1-2/+9
The driver_override implementation is susceptible to a race condition when different threads are reading vs. storing a different driver override. Add locking to avoid the race condition. This is in close analogy to commit 6265539776a0 ("driver core: platform: fix race condition with driver_override") from Adrian Salido. Fixes: 782a985d7af2 ("PCI: Introduce new device binding path using pci_dev.driver_override") Signed-off-by: Nicolai Stange <[email protected]> Signed-off-by: Bjorn Helgaas <[email protected]> Cc: [email protected] # v3.16+
2017-09-25net: qcom/emac: specify the correct size when mapping a DMA bufferTimur Tabi1-1/+2
When mapping the RX DMA buffers, the driver was accidentally specifying zero for the buffer length. Under normal circumstances, SWIOTLB does not need to allocate a bounce buffer, so the address is just mapped without checking the size field. This is why the error was not detected earlier. Fixes: b9b17debc69d ("net: emac: emac gigabit ethernet controller driver") Cc: [email protected] Signed-off-by: Timur Tabi <[email protected]> Signed-off-by: David S. Miller <[email protected]>
2017-09-26cpufreq: dt: Fix sysfs duplicate filename creation for platform-deviceSuniel Mahesh1-0/+4
ti-cpufreq and cpufreq-dt-platdev drivers are registering platform-device with same name "cpufreq-dt" using platform_device_register_*() routines. This is leading to build warnings appended below. Providing hardware information to OPP framework along with the platform- device creation should be done by ti-cpufreq driver before cpufreq-dt driver comes into place. This patch add's TI am33xx, am43 and dra7 platforms (which use opp-v2 property) to the blacklist of devices in cpufreq-dt-platform driver to avoid creating platform-device twice and remove build warnings. [ 2.370167] ------------[ cut here ]------------ [ 2.375087] WARNING: CPU: 0 PID: 1 at fs/sysfs/dir.c:31 sysfs_warn_dup+0x58/0x78 [ 2.383112] sysfs: cannot create duplicate filename '/devices/platform/cpufreq-dt' [ 2.391219] Modules linked in: [ 2.394506] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.13.0-next-20170912 #1 [ 2.402006] Hardware name: Generic AM33XX (Flattened Device Tree) [ 2.408437] [<c0110a28>] (unwind_backtrace) from [<c010ca84>] (show_stack+0x10/0x14) [ 2.416568] [<c010ca84>] (show_stack) from [<c0827d64>] (dump_stack+0xac/0xe0) [ 2.424165] [<c0827d64>] (dump_stack) from [<c0137470>] (__warn+0xd8/0x104) [ 2.431488] [<c0137470>] (__warn) from [<c01374d0>] (warn_slowpath_fmt+0x34/0x44) [ 2.439351] [<c01374d0>] (warn_slowpath_fmt) from [<c03459d0>] (sysfs_warn_dup+0x58/0x78) [ 2.447938] [<c03459d0>] (sysfs_warn_dup) from [<c0345ab8>] (sysfs_create_dir_ns+0x80/0x98) [ 2.456719] [<c0345ab8>] (sysfs_create_dir_ns) from [<c082c554>] (kobject_add_internal+0x9c/0x2d4) [ 2.466124] [<c082c554>] (kobject_add_internal) from [<c082c7d8>] (kobject_add+0x4c/0x9c) [ 2.474712] [<c082c7d8>] (kobject_add) from [<c05803e4>] (device_add+0xcc/0x57c) [ 2.482489] [<c05803e4>] (device_add) from [<c0584b74>] (platform_device_add+0x100/0x220) [ 2.491085] [<c0584b74>] (platform_device_add) from [<c05855a8>] (platform_device_register_full+0xf4/0x118) [ 2.501305] [<c05855a8>] (platform_device_register_full) from [<c067023c>] (ti_cpufreq_init+0x150/0x22c) [ 2.511253] [<c067023c>] (ti_cpufreq_init) from [<c0101df4>] (do_one_initcall+0x3c/0x170) [ 2.519838] [<c0101df4>] (do_one_initcall) from [<c0c00eb4>] (kernel_init_freeable+0x1fc/0x2c4) [ 2.528974] [<c0c00eb4>] (kernel_init_freeable) from [<c083bcac>] (kernel_init+0x8/0x110) [ 2.537565] [<c083bcac>] (kernel_init) from [<c0107d18>] (ret_from_fork+0x14/0x3c) [ 2.545981] ---[ end trace 2fc00e213c13ab20 ]--- [ 2.551051] ------------[ cut here ]------------ [ 2.555931] WARNING: CPU: 0 PID: 1 at lib/kobject.c:240 kobject_add_internal+0x254/0x2d4 [ 2.564578] kobject_add_internal failed for cpufreq-dt with -EEXIST, don't try to register things with the same name in the same directory. [ 2.577977] Modules linked in: [ 2.581261] CPU: 0 PID: 1 Comm: swapper/0 Tainted: G W 4.13.0-next-20170912 #1 [ 2.590013] Hardware name: Generic AM33XX (Flattened Device Tree) [ 2.596437] [<c0110a28>] (unwind_backtrace) from [<c010ca84>] (show_stack+0x10/0x14) [ 2.604573] [<c010ca84>] (show_stack) from [<c0827d64>] (dump_stack+0xac/0xe0) [ 2.612172] [<c0827d64>] (dump_stack) from [<c0137470>] (__warn+0xd8/0x104) [ 2.619494] [<c0137470>] (__warn) from [<c01374d0>] (warn_slowpath_fmt+0x34/0x44) [ 2.627362] [<c01374d0>] (warn_slowpath_fmt) from [<c082c70c>] (kobject_add_internal+0x254/0x2d4) [ 2.636666] [<c082c70c>] (kobject_add_internal) from [<c082c7d8>] (kobject_add+0x4c/0x9c) [ 2.645255] [<c082c7d8>] (kobject_add) from [<c05803e4>] (device_add+0xcc/0x57c) [ 2.653027] [<c05803e4>] (device_add) from [<c0584b74>] (platform_device_add+0x100/0x220) [ 2.661615] [<c0584b74>] (platform_device_add) from [<c05855a8>] (platform_device_register_full+0xf4/0x118) [ 2.671833] [<c05855a8>] (platform_device_register_full) from [<c067023c>] (ti_cpufreq_init+0x150/0x22c) [ 2.681779] [<c067023c>] (ti_cpufreq_init) from [<c0101df4>] (do_one_initcall+0x3c/0x170) [ 2.690377] [<c0101df4>] (do_one_initcall) from [<c0c00eb4>] (kernel_init_freeable+0x1fc/0x2c4) [ 2.699510] [<c0c00eb4>] (kernel_init_freeable) from [<c083bcac>] (kernel_init+0x8/0x110) [ 2.708106] [<c083bcac>] (kernel_init) from [<c0107d18>] (ret_from_fork+0x14/0x3c) [ 2.716217] ---[ end trace 2fc00e213c13ab21 ]--- Fixes: edeec420de24 (cpufreq: dt-cpufreq: platdev Automatically create device with OPP v2) Signed-off-by: Suniel Mahesh <[email protected]> Acked-by: Viresh Kumar <[email protected]> Signed-off-by: Rafael J. Wysocki <[email protected]>
2017-09-25scsi: scsi_transport_fc: set scsi_target_id upon rescanHannes Reinecke1-10/+1
When an rport is found in the bindings array there is no guarantee that it had been a target port, so we need to call fc_remote_port_rolechg() here to ensure the scsi_target_id is set correctly. Otherwise the port will never be scanned. Signed-off-by: Hannes Reinecke <[email protected]> Reviewed-by: Johannes Thumshirn <[email protected]> Tested-by: Chad Dupuis <[email protected]> Signed-off-by: Martin K. Petersen <[email protected]>
2017-09-25Merge branch 'for-linus' of git://git.kernel.dk/linux-blockLinus Torvalds15-128/+128
Pull block fixes from Jens Axboe: - Two sets of NVMe pull requests from Christoph: - Fixes for the Fibre Channel host/target to fix spec compliance - Allow a zero keep alive timeout - Make the debug printk for broken SGLs work better - Fix queue zeroing during initialization - Set of RDMA and FC fixes - Target div-by-zero fix - bsg double-free fix. - ndb unknown ioctl fix from Josef. - Buffered vs O_DIRECT page cache inconsistency fix. Has been floating around for a long time, well reviewed. From Lukas. - brd overflow fix from Mikulas. - Fix for a loop regression in this merge window, where using a union for two members of the loop_cmd turned out to be a really bad idea. From Omar. - Fix for an iostat regression fix in this series, using the wrong API to get at the block queue. From Shaohua. - Fix for a potential blktrace delection deadlock. From Waiman. * 'for-linus' of git://git.kernel.dk/linux-block: (30 commits) nvme-fcloop: fix port deletes and callbacks nvmet-fc: sync header templates with comments nvmet-fc: ensure target queue id within range. nvmet-fc: on port remove call put outside lock nvme-rdma: don't fully stop the controller in error recovery nvme-rdma: give up reconnect if state change fails nvme-core: Use nvme_wq to queue async events and fw activation nvme: fix sqhd reference when admin queue connect fails block: fix a crash caused by wrong API fs: Fix page cache inconsistency when mixing buffered and AIO DIO nvmet: implement valid sqhd values in completions nvme-fabrics: Allow 0 as KATO value nvme: allow timed-out ios to retry nvme: stop aer posting if controller state not live nvme-pci: Print invalid SGL only once nvme-pci: initialize queue memory before interrupts nvmet-fc: fix failing max io queue connections nvme-fc: use transport-specific sgl format nvme: add transport SGL definitions nvme.h: remove FC transport-specific error values ...
2017-09-26PM / OPP: Call notifier without holding opp_table->lockViresh Kumar1-0/+7
The notifier callbacks may want to call some OPP helper routines which may try to take the same opp_table->lock again and cause a deadlock. One such usecase was reported by Chanwoo Choi, where calling dev_pm_opp_disable() leads us to the devfreq's OPP notifier handler, which further calls dev_pm_opp_find_freq_floor() and it deadlocks. We don't really need the opp_table->lock to be held across the notifier call though, all we want to make sure is that the 'opp' doesn't get freed while being used from within the notifier chain. We can do it with help of dev_pm_opp_get/put() as well. Let's do it. Cc: 4.11+ <[email protected]> # 4.11+ Fixes: 5b650b388844 "PM / OPP: Take kref from _find_opp_table()" Reported-by: Chanwoo Choi <[email protected]> Tested-by: Chanwoo Choi <[email protected]> Reviewed-by: Stephen Boyd <[email protected]> Reviewed-by: Chanwoo Choi <[email protected]> Signed-off-by: Viresh Kumar <[email protected]> Signed-off-by: Rafael J. Wysocki <[email protected]>
2017-09-25scsi: scsi_transport_iscsi: fix the issue that iscsi_if_rx doesn't parse ↵Xin Long1-1/+1
nlmsg properly ChunYu found a kernel crash by syzkaller: [ 651.617875] kasan: CONFIG_KASAN_INLINE enabled [ 651.618217] kasan: GPF could be caused by NULL-ptr deref or user memory access [ 651.618731] general protection fault: 0000 [#1] SMP KASAN [ 651.621543] CPU: 1 PID: 9539 Comm: scsi Not tainted 4.11.0.cov #32 [ 651.621938] Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011 [ 651.622309] task: ffff880117780000 task.stack: ffff8800a3188000 [ 651.622762] RIP: 0010:skb_release_data+0x26c/0x590 [...] [ 651.627260] Call Trace: [ 651.629156] skb_release_all+0x4f/0x60 [ 651.629450] consume_skb+0x1a5/0x600 [ 651.630705] netlink_unicast+0x505/0x720 [ 651.632345] netlink_sendmsg+0xab2/0xe70 [ 651.633704] sock_sendmsg+0xcf/0x110 [ 651.633942] ___sys_sendmsg+0x833/0x980 [ 651.637117] __sys_sendmsg+0xf3/0x240 [ 651.638820] SyS_sendmsg+0x32/0x50 [ 651.639048] entry_SYSCALL_64_fastpath+0x1f/0xc2 It's caused by skb_shared_info at the end of sk_buff was overwritten by ISCSI_KEVENT_IF_ERROR when parsing nlmsg info from skb in iscsi_if_rx. During the loop if skb->len == nlh->nlmsg_len and both are sizeof(*nlh), ev = nlmsg_data(nlh) will acutally get skb_shinfo(SKB) instead and set a new value to skb_shinfo(SKB)->nr_frags by ev->type. This patch is to fix it by checking nlh->nlmsg_len properly there to avoid over accessing sk_buff. Reported-by: ChunYu Wang <[email protected]> Signed-off-by: Xin Long <[email protected]> Acked-by: Chris Leech <[email protected]> Signed-off-by: Martin K. Petersen <[email protected]>
2017-09-25irqchip/mips-gic: Use effective affinity to unmaskPaul Burton1-2/+5
Commit 7778c4b27cbe ("irqchip: mips-gic: Use pcpu_masks to avoid reading GIC_SH_MASK*") adjusted the way we handle masking interrupts to set & clear the interrupt's bit in each pcpu_mask. This allows us to avoid needing to read the GIC mask registers and perform a bitwise and of their values with the pending & pcpu_masks. Unfortunately this didn't quite work for IPIs, which were mapped to a particular CPU/VP during initialisation but never set the affinity or effective_affinity fields of their struct irq_desc. This led to them losing their affinity when gic_unmask_irq() was called for them, and they'd all become affine to cpu0. Fix this by: 1) Setting the effective affinity of interrupts in gic_shared_irq_domain_map(), which is where we actually map an interrupt to a CPU/VP. This ensures that the effective affinity mask is always valid, not just after explicitly setting affinity. 2) Using an interrupt's effective affinity when unmasking it, which prevents gic_unmask_irq() from unintentionally changing which pcpu_mask includes an interrupt. Fixes: 7778c4b27cbe ("irqchip: mips-gic: Use pcpu_masks to avoid reading GIC_SH_MASK*") Signed-off-by: Paul Burton <[email protected]> Signed-off-by: Thomas Gleixner <[email protected]> Cc: Marc Zyngier <[email protected]> Cc: Jason Cooper <[email protected]> Link: https://lkml.kernel.org/r/[email protected]
2017-09-25irqchip/mips-gic: Fix shifts to extract register fieldsPaul Burton1-3/+3
The MIPS GIC driver is incorrectly using __fls to shift registers, intending to shift to the least significant bit of a value based upon its mask but instead shifting off all but the value's top bit. It should actually be using __ffs to shift to the first, not last, bit of the value. Apparently the system I used when testing commit 3680746abd87 ("irqchip: mips-gic: Convert remaining shared reg access to new accessors") and commit b2b2e584ceab ("irqchip: mips-gic: Clean up mti, reserved-cpu-vectors handling") managed to work correctly despite this issue, but not all systems do... Fixes: 3680746abd87 ("irqchip: mips-gic: Convert remaining shared reg access to new accessors") Fixes: b2b2e584ceab ("irqchip: mips-gic: Clean up mti, reserved-cpu-vectors handling") Signed-off-by: Paul Burton <[email protected]> Signed-off-by: Thomas Gleixner <[email protected]> Cc: Marc Zyngier <[email protected]> Cc: Jason Cooper <[email protected]> Link: https://lkml.kernel.org/r/[email protected]
2017-09-25nvme-fcloop: fix port deletes and callbacksJames Smart1-64/+38
Now that there are potentially long delays between when a remoteport or targetport delete calls is made and when the callback occurs (dev_loss_tmo timeout), no longer block in the delete routines and move the final nport puts to the callbacks. Moved the fcloop_nport_get/put/free routines to avoid forward declarations. Ensure port_info structs used in registrations are nulled in case fields are not set (ex: devloss_tmo values). Signed-off-by: James Smart <[email protected]> Signed-off-by: Christoph Hellwig <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
2017-09-25nvmet-fc: ensure target queue id within range.James Smart1-0/+3
When searching for queue id's ensure they are within the expected range. Signed-off-by: James Smart <[email protected]> Signed-off-by: Christoph Hellwig <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
2017-09-25nvmet-fc: on port remove call put outside lockJames Smart1-1/+5
Avoid calling the put routine, as it may traverse to free routines while holding the target lock. Signed-off-by: James Smart <[email protected]> Signed-off-by: Christoph Hellwig <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
2017-09-25nvme-rdma: don't fully stop the controller in error recoverySagi Grimberg1-1/+1
By calling nvme_stop_ctrl on a already failed controller will wait for the scan work to complete (only by identify timeout expiration which is 60 seconds). This is unnecessary when we already know that the controller has failed. Reported-by: Yi Zhang <[email protected]> Signed-off-by: Sagi Grimberg <[email protected]> Signed-off-by: Christoph Hellwig <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
Powered by cgit, Gruvboxed by Blaster4385.