- 18 Sep, 2016 14 commits
-
-
Linus Walleij authored
This adds support for the mounting matrix to the KXSD9 driver. Signed-off-by:
Linus Walleij <linus.walleij@linaro.org> Signed-off-by:
Jonathan Cameron <jic23@kernel.org>
-
Linus Walleij authored
This deploys runtime and system PM in the KXSD9 driver: - Use the force_runtime_suspend/resume callbacks as system PM operations. - Add buffer prepare/unprepare callbacks to grab the runtime PM while we're using buffered reads and put get/put_autosuspend in these. - Insert get/put_autosuspend calls anywhere the IO is used from the raw read/write callbacks. - Move the fullscale setting to be cached in the state container so we can restore it properly when coming back from system/runtime suspend. - Set the autosuspend delay to two orders of magnitude that of the sensor start-up time (20ms) so we will autosuspend after 2s. - Register the callbacks in both the SPI and I2C subdrivers. Tested with the I2C KXSD9 on the Qualcomm APQ8060 Dragonboard. Signed-off-by:
-
Linus Walleij authored
What is passed to the .probe() and .remove() functions is technically the parent of the created IIO device but it becomes a big confusion for the head to have it named like this since it is usually clear from context the "dev" refers to the physical device, and when next adding PM callbacks a clean "struct device *dev" pointer is passed to these and that makes it even more confused. Rename "parent" to "dev" like in most other drivers. Signed-off-by:
-
Linus Walleij authored
This adds supply regulator handling for the VDD and IOVDD inputs on the KXSD9 component, makes sure to bring the regulators online during probe and disable them on remove or the errorpath. Signed-off-by:
-
Linus Walleij authored
There are some hardcoded register values etc in the code, define proper bitfield definitions, and use them when getting and setting the scale. Optimize a read/modify/write to use regmap_update_bits() at the same time. Signed-off-by:
-
Linus Walleij authored
As is custom with all modern sensors, add a clever burst mode that will just stream out values from the sensor and provide it to userspace to do the proper offsetting and scaling. This is the result when tested with an HRTimer trigger: $ generic_buffer -a -c 10 -n kxsd9 -t foo /sys/bus/iio/devices/iio:device1 foo 0.371318 0.718680 9.869872 1795.000000 97545896129 -0.586922 0.179670 9.378775 2398.000000 97555864721 -0.299450 0.179670 10.348992 2672.000000 97565874055 0.371318 0.335384 11.103606 2816.000000 97575883240 0.179670 0.574944 10.540640 2847.000000 97585862351 0.335384 0.754614 9.953718 2840.000000 97595872425 0.179670 0.754614 10.732288 2879.000000 97605882351 0.000000 0.754614 10.348992 2872.000000 97615891832 -0.730658 0.574944 9.570422 2831.000000 97625871536 0.000000 1.137910 10.732288 2872.000000 97635881610 Columns shown are x, y, z acceleration, so a positive acceleration of ~9.81 (shaky due to bad calibration) along the z axis. The fourth column is the AUX IN which is floating on this system, it seems to float up to the 2.85V VDD voltage. To be able to cleanup the triggered buffer, we need to add .remove() callbacks to the I2C and SPI subdrivers and call back into an exported .remove() callback in the core. Signed-off-by:
-
Linus Walleij authored
This fixes several errors in the offset and scaling of the raw values from the KXSD9 sensor: - The code did not convert the big endian value from the sensor into the endianness of the host CPU. Fix this with be16_to_cpu() on the raw obtained value. - The code did not regard the fact that only the upper 12 bits of the accelerometer values are valid. Shift these down four bits to yield the real raw value. - Further the sensor provides 2048 at zero g. This means that an offset of 2048 must be subtracted from the raw value before scaling. This was not taken into account by the driver, yielding a weird value. Fix this by providing this offset in sysfs. To house the scaling code better, the value reading code was factored into the raw reading function. This proper scaling and offseting is necessary to get proper values out of triggered buffer by offsetting, shifting and scaling them. Signed-off-by:
-
Linus Walleij authored
The RX/TX buffers are gone so drop the lock (it should have been in the transport struct anyway). Tested-by:
-
Linus Walleij authored
This adds I2C regmap transport for the KXSD9 driver. Tested on the KXSD9 sensor on the APQ8060 Dragonboard. Signed-off-by:
-
Linus Walleij authored
This converts the KXSD9 driver to drop the custom transport mechanism and just use regmap like everything else. Tested-by:
-
Linus Walleij authored
This is just a masquerading register write function, so use the register write function instead. Signed-off-by:
-
Linus Walleij authored
This moves the KXSD9 SPI transport out to its own file and Kconfig entry, so that we will be able to add another transport method. We export the common probe and add a local header file for the functionality shared between the main driver and the transport driver. We make the SPI transport the default for the driver if SPI is available and the KXSD9 driver was selected, so the oldconfig upgrade path will be clear. Signed-off-by:
-
Linus Walleij authored
This makes it possible to later split the transport mechanism using a generic probe() and a generic remove(). Use dev_set_drvdata() and dev_get_drvdata() as a paired accessor to operate on the abstract struct device * regardless of the transport mechanism in use. Signed-off-by:
-
Linus Walleij authored
Split off a transport mechanism struct that will deal with the SPI traffic in preparation for adding I2C support. Tested-by:
-
- 17 Sep, 2016 20 commits
-
-
Cathal Mullaney authored
Fixed a coding style issue. Signed-off-by:
Cathal Mullaney <chuckleberryfinn@gmail.com> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Namrata A Shettar authored
Correct the word 'success' being misspelt in the macro: - LED_BLINK_WPS_SUCESS_INTERVAL_ALPHA => LED_BLINK_WPS_SUCCESS_INTERVAL_ALPHA Signed-off-by:
Namrata A Shettar <namrataashettar@gmail.com> Signed-off-by:
-
sayli karnik authored
staging: rtl8188eu: core: Replace memcpy() with ether_addr_copy() if ethernet addresses are __aligned(2) This patch fixes the checkpatch warning: Prefer ether_addr_copy() over memcpy() if the Ethernet addresses are __aligned(2) Done using coccinelle: @@ expression exp1, exp2; @@ - memcpy(exp1, exp2, ETH_ALEN); + ether_addr_copy(exp1, exp2); The output of pahole shows that the addresses are aligned: struct ieee80211_hdr { __le16 frame_control; /* 0 2 */ __le16 duration_id; /* 2 2 */ u8 addr1[6]; /* 4 6 */ u8 addr2[6]; /* 10 6 */ u8 addr3[6]; /* 16 6 */ __le16 seq_ctrl; /* 22 2 */ u8 addr4[6]; /* 24 6 */ /* size: 30, cachelines: 1, members: 7 */ /* last cacheline: 30 bytes */ }; struct wlan_bssid_ex { u32 Length; /* 0 4 */ unsigned char MacAddress[6]; /* 4 6 */ u8 Reserved[2]; /* 10 2 */ struct ndis_802_11_ssid Ssid; /* 12 36 */ u32 Privacy; /* 48 4 */ long int Rssi; /* 52 8 */ enum NDIS_802_11_NETWORK_TYPE NetworkTypeInUse; /* 60 4 */ /* --- cacheline 1 boundary (64 bytes) --- */ struct ndis_802_11_config Configuration; /* 64 32 */ enum ndis_802_11_network_infra InfrastructureMode; /* 96 4 */ unsigned char SupportedRates[16]; /* 100 16 */ struct wlan_phy_info PhyInfo; /* 116 4 */ u32 IELength; /* 120 4 */ u8 IEs[768]; /* 124 768 */ /* --- cacheline 13 boundary (832 bytes) was 60 bytes ago --- */ /* size: 892, cachelines: 14, members: 13 */ /* last cacheline: 60 bytes */ }; struct stadel_event { unsigned char macaddr[6]; /* 0 6 */ unsigned char rsvd[2]; /* 6 2 */ int mac_id; /* 8 4 */ /* size: 12, cachelines: 1, members: 3 */ /* last cacheline: 12 bytes */ }; Signed-off-by:sayli karnik <karniksayli1995@gmail.com> Signed-off-by:
-
Andrew Kanner authored
Ran checkpatch.pl -f vme_pio2_core.c Fixed: WARNING: Symbolic permissions are not preferred. Consider using octal permissions (0444) Signed-off-by:
Andrew Kanner <andrew.kanner@gmail.com> Signed-off-by:
-
sayli karnik authored
This patch replaces type u_int8_t with u8 as suggested by checkpatch.pl. Signed-off-by:
-
sayli karnik authored
This patch replaces the explicit NULL comparisons with ! operator. It was found using Coccinelle: @disable is_null@ expression e; @@ - e == NULL + !e Signed-off-by:
Julia Lawall <julia.lawall@lip6.fr> Signed-off-by:
-
sayli karnik authored
This patch replaces left shifts on 1 with the BIT(x) macro, as suggested by checkpatch.pl. This was done using coccinelle script: @@ constant c; @@ - (1 << c) + BIT(c) Signed-off-by:
-
Bhumika Goyal authored
Remove extern specifier from function declaration as they have it by default. Also move extern declaration from .c files to their respective header file 'ks_hostif.h'. Coccinelle was used to remove extern and other changes were done by hand. Script: @@ identifier func; type T; @@ - extern T func(...); Signed-off-by:
Bhumika Goyal <bhumirks@gmail.com> Signed-off-by:
-
Namrata A Shettar authored
Add space around binary operators to resolve checkpatch issue. Signed-off-by:
-
Baoyou Xie authored
We get a few warnings when building kernel with W=1: drivers/staging/lustre/lustre/lmv/lmv_obd.c:1640:1: warning: no previous prototype for 'lmv_locate_target_for_name' [-Wmissing-prototypes] drivers/staging/lustre/lustre/lmv/lmv_obd.c:2421:5: warning: no previous prototype for 'lmv_read_page' [-Wmissing-prototypes] .... In fact, these functions are only used in the file in which they are declared and don't need a declaration, but can be made static. so this patch marks these functions with 'static'. Signed-off-by:
Baoyou Xie <baoyou.xie@linaro.org> Signed-off-by:
-
Baoyou Xie authored
We get 1 warning when building kernel with W=1: drivers/staging/lustre/lustre/obdclass/cl_page.c:488:6: warning: no previous prototype for 'cl_page_disown0' [-Wmissing-prototypes] In fact, this function is declared in drivers/staging/lustre/lustre/obdclass/cl_io.c, but should be removed in a header file. thus can be recognized in other file. So this patch moves the declaration into drivers/staging/lustre/lustre/include/cl_object.h. Signed-off-by:
-
Baoyou Xie authored
We get 3 warnings when building kernel with W=1: drivers/staging/lustre/lustre/obdclass/obd_mount.c:705:5: warning: no previous prototype for 'lustre_check_exclusion' [-Wmissing-prototypes] drivers/staging/lustre/lustre/obdclass/obd_mount.c:1226:5: warning: no previous prototype for 'lustre_register_fs' [-Wmissing-prototypes] drivers/staging/lustre/lustre/obdclass/obd_mount.c:1231:5: warning: no previous prototype for 'lustre_unregister_fs' [-Wmissing-prototypes] In fact, the first function is declared in drivers/staging/lustre/lustre/obdclass/obd_config.c, and the other two functions are declared in drivers/staging/lustre/lustre/obdclass/class_obd.c, but all of these functions should be declared in a header file, thus can be recognized in other file. So this patch moves the declarations into drivers/staging/lustre/lustre/include/obd_class.h. Signed-off-by:
-
Baoyou Xie authored
We get 1 warning when building kernel with W=1: drivers/staging/lustre/lustre/obdclass/obd_config.c:719:6: warning: no previous prototype for 'lustre_register_client_process_config' [-Wmissing-prototypes] In fact, this function is declared in drivers/staging/lustre/lustre/llite/super25.c, but should be moved into a header file, thus can be recognized in other file. So this patch moves the declaration into drivers/staging/lustre/lustre/include/lprocfs_status.h. Signed-off-by:
-
Baoyou Xie authored
We get 1 warning when building kernel with W=1: drivers/staging/lustre/lnet/selftest/conctl.c:804:1: warning: no previous prototype for 'lstcon_ioctl_entry' [-Wmissing-prototypes] In fact, this function is declared in drivers/staging/lustre/lustre/llite/super25.c, but should be moved into a header file, thus can be recognized in other file. So this patch moves the declaration into drivers/staging/lustre/lnet/selftest/console.h Signed-off-by:
-
Richard Groux authored
Minor warnings spotted by checkpatch.pl in lustre Remove unnecessary space before function pointer arguments. Signed-off-by:
Richard Groux <rgroux@sauron-mordor.net> Signed-off-by:
-
Bhumika Goyal authored
Use eth_zero_addr to assign zero address to the given address array instead of memset when the second argument in memset is address of zero. Coccinelle was used to do the replacement and add the header file linux/etherdevice.h if not already present. Script: @header@ @@ #include <linux/etherdevice.h> @r1@ expression e; @@ -memset(e,0,ETH_ALEN); +eth_zero_addr(e); @includeheader depends on r1 && !header@ @@ + #include <linux/etherdevice.h> #include <...> Signed-off-by:
-
sayli karnik authored
The patch uses the size of the dereference to a variable instead of it's type, to make the code more resistant to type changes in the future. The type of the pointer variable here is the same as the type in the argument that is being replaced in sizeof(). Signed-off-by:
-
Ivan Safonov authored
This function used once, and does not improve code readability. Signed-off-by:
Ivan Safonov <insafonov@gmail.com> Signed-off-by:
-
Ivan Safonov authored
This function used once, and does not improve code readability. Signed-off-by:
-
Ivan Safonov authored
This function does not improve code readability. Signed-off-by:
-
- 16 Sep, 2016 6 commits
-
-
Gargi Sharma authored
add spaces around << to fix the checkpatch issue, spaces preferred around that '<<'. Signed-off-by:
Gargi Sharma <gs051095@gmail.com> Signed-off-by:
-
Gargi Sharma authored
add spaces around | to fix the checkpatch issue, spaces preferred around that '|'. Signed-off-by:
-
Gargi Sharma authored
add spaces around >> to fix the checkpatch issue, spaces preferred around that '>>'. Signed-off-by:
-
Gargi Sharma authored
add space around & to fix the checkpatch issue spaces preferred around that '&'. Signed-off-by:
-
Gargi Sharma authored
add spaces around + to fix the checkpatch issue, spaces preferred around that '&'. Signed-off-by:
-
Gargi Sharma authored
add spaces around << to fix the checkpatch issue, spaces preferred around that '<<'. Signed-off-by:
-
