September 2018 - kernelnewbies@kernelnewbies.org

4.19-rc3 incompatible with nvidia driver
by Daan Wendelen Sept. 21, 2018

Sept. 21, 2018

Hi, The latest nvidia driver is incompatible with 4.19-rc3 because "drm_mode_connector_update_edid_property" got renamed to "drm_connector_update_edid_property" and "drm_mode_connector_attach_encoder" to "drm_connector_attach_encoder" in the linux headers. Is this considered the fault of linux or is it considered nvidia's responsibility to fix this? If it's considered regression in the Linux kernel, to who (/ which mailing list) do I report this? Daan

3 4

regarding const variables/structures
by inventsekar Sept. 18, 2018

Sept. 18, 2018

Hi All... One curious question.. Linux Foundation tweeted this - Meet Bhumika Goyal, age 22, from India. She has had more 340 patches accepted into the Linux kernel, which helped her land one of our two Linux Kernel Guru LiFT scholarships: https://twitter.com/linuxfoundation/status/940340927897489408?lang=en and her commits are - https://github.com/torvalds/linux/commits?author=bhumikagoyal One of my friend told me that, most/many of patches are just converting variables to "*constant*" For example - platform/chrome: chromeos_laptop: make chromeos_laptop const - Declare chromeos_laptop structures as *const *as they are only used during a copy operation. As their value is never modified during runtime, they can be made const. the question is that, how these many const variable issues are left/missed by the previous developers?!?! per my little knowledge, this task looks like a simple one.. many other Kernel Developers should have thought and made these changes long before, but i am not able to understand why this wasnt done until recently?!?!? -- Best Regards, Sekar

7 14

IOCtl argument check
by Subhashini Rao Beerisetty Sept. 17, 2018

Sept. 17, 2018

Hi All, Can someone clarify me on IOCtl argument check. Kernel code has few IOCtl’s with a pointer to structure as an arguments. For example please consider the below mentioned IOCtl’s. #define MB862XX_L1_SET_CFG _IOW(MB862XX_BASE, 1, struct mb862xx_l1_cfg*) #define MB862XX_L1_GET_CFG _IOR(MB862XX_BASE, 0, struct mb862xx_l1_cfg*) #define AGPIOC_INFO _IOR (AGPIOC_BASE, 0, struct agp_info*) If user space passes a structure object instead of pointer to struct, what should be the return value from kernel code? How to debug this kind scenario’s if copy_{to,from}_user returns success(i.e. “0”)? The IOCtl’s are represented in 32-bits. Out of these 13 bits(bits 29-16) are reserved for size of argument. If user space passes an object instead of pointer and assume size of the structure is not equal to size of the pointer, then IOCtl decoding should result a different value, right? So for this case, does user space IOCtl calls the corresponding driver IOCtl function? Thanks

1 0

interruptible sleep tasks
by Muni Sekhar Sept. 11, 2018

Sept. 11, 2018

Hi All, During kernel drivers testing I see lot of user mode applications blocks forever in interruptible sleep mode. Is there any way to get the corresponding kernel functions stack trace ? I see kernel has “hung_task_timeout_secs” parameter and when a task in D state did not get scheduled for more than this value report a warning. Similarly is it possible to get a warning or kernel functions stack trace when a task in “S - interruptible sleep mode” and did not get scheduled for certain duration? -- Thanks, Sekhar

2 1

Build using O=path
by Tobin C. Harding Sept. 10, 2018

Sept. 10, 2018

Hi, I'm unable to build the kernel using the O=path/to/out option. Am I doing something brain dead? $ cd $KERNEL $ make mrproper $ cp path/to/valid/config/file .config $ make O=/abs/path/to/output ... scripts/kconfig/conf --syncconfig Kconfig *** *** Configuration file ".config" not found! *** *** Please run some configurator (e.g. "make oldconfig" or *** "make menuconfig" or "make xconfig"). ... So I tried copying the .config file to the output directory $ cp .config /abs/path/to/output $ make O=/abs/path/to/output ... Using /home/tobin/build/kernel/linux as source for kernel /home/tobin/build/kernel/linux is not clean, please run 'make mrproper' in the '/home/tobin/build/kernel/linux' directory. /home/tobin/build/kernel/linux/Makefile:1091: recipe for target 'prepare3' failed make[1]: *** [prepare3] Error 1 make[1]: Leaving directory '/home/tobin/build/kernel/output/master' Makefile:146: recipe for target 'sub-make' failed make: *** [sub-make] Error 2 Same thing if I use KBUILD_OUTPUT instead of passing O= thanks, Tobin.

4 5

gdb debug get_unmapped_area
by Dengke Du Sept. 10, 2018

Sept. 10, 2018

Hi When I use gdb debug get_unmapped_area on qemumips64, I found a strange phenomenon: file: mm/mmap.c function: get_unmapped_area line: addr = get_area(file, addr, len, pgoff, flags); This return -22, I want to find where return -22, then I step in get_area: (gdb) s shmem_get_unmapped_area (file=0x900000000888b000, uaddr=18446744073709547520, len=4096, pgoff=0, flags=17) at /usr/src/kernel/mm/shmem.c:2043 2043 /usr/src/kernel/mm/shmem.c: No such file or directory. (gdb) n arch_get_unmapped_area_topdown (filp=0x900000000888b000, addr0=18446744073709547520, len=4096, pgoff=0, flags=17) at /usr/src/kernel/arch/mips/mm/mmap.c:142 142 /usr/src/kernel/arch/mips/mm/mmap.c: No such file or directory. (gdb) n 143 in /usr/src/kernel/arch/mips/mm/mmap.c (gdb) n arch_get_unmapped_area_common (filp=0x900000000888b000, addr0=18446744073709547520, len=4096, pgoff=0, flags=17, dir=DOWN) at /usr/src/kernel/arch/mips/mm/mmap.c:59 59 in /usr/src/kernel/arch/mips/mm/mmap.c (gdb) n 66 in /usr/src/kernel/arch/mips/mm/mmap.c Why it jump three three function? shmem_get_unmapped_area arch_get_unmapped_area_topdown arch_get_unmapped_area_common

1 0

ATA link errors in custom PATA driver
by Nathan Williams Sept. 5, 2018

Sept. 5, 2018

Hi, I have a Altera socfpga platform that implements a very basic ATA PIO0 Compact Flash controller in the FPGA and I have a very basic PATA Linux driver that does MMIO to the FPGA registers and configures 3 GPIOs for power, reset and card detect. It's basically a copy of pata_platform.c with a check_status function to read the card detect GPIO and if a card is detected, it switches on the power GPIO and deasserts the reset GPIO. The driver and FPGA CF controller work fine together for me except for an intermittent issue where on boot I occasionally see the following: [ 5.836242] ata1: link is slow to respond, please be patient (ready=0) [ 10.816241] ata1: device not ready (errno=-16), forcing hardreset [ 16.016250] ata1: link is slow to respond, please be patient (ready=0) [ 20.876239] ata1: SRST failed (errno=-16) [ 26.076243] ata1: link is slow to respond, please be patient (ready=0) [ 30.936243] ata1: SRST failed (errno=-16) [ 36.136243] ata1: link is slow to respond, please be patient (ready=0) [ 65.956240] ata1: SRST failed (errno=-16) [ 70.976238] ata1: SRST failed (errno=-16) [ 70.981059] ata1: reset failed, giving up This is occurs when there is no CF card plugged in, so I don't understand why the driver appears to be trying to communicate with a Compact Flash device? I'm a little stuck trying to figure out how to properly handle a hot pluggable CF card, so any advice on this would be much appreciated. Nathan

1 0

Implementing tight time constraints for read/write with USB serial (ftdi_sio)
by Thomas Bracht Laumann Jespersen Sept. 5, 2018

Sept. 5, 2018

Hello all! I'm working on a project where we want to integrate into a production line and listen on connected devices, and the challenge is that our setup needs to be fast enough. Let me try to explain the setup. There are some 80+ devices on the network. They are connected in a multidrop talking PROFIBUS at a speed of 1.5MB. I have an RS485<->USB serial cable that uses the ftdi_sio driver to listen on the network. It is identified as a FT232RL. Just listening and de-constructing the PROFIBUS communication blocks that are coming through is not a problem. The problem is when I need to reply to messages in the network. At first I was seeing reads from the serial port taking either ~.002ms or ~16ms, which I assume is due to buffering on the chip. I then learned about setserial and tried adjusting the latency to 1ms. This gives me read times at around ~4ms on average (but sometimes going above 10ms). I'm looking for ways to disable/minimize the buffering on the FTDI chip, but I also suspect that the ftdi_sio driver does some buffering. The challenge is that my application needs to act within 100 tbits, ie in around ~.066ms. That is the window in which I need to begin transmitting a reply. Is there a way to achieve this with the ftdi_sio driver or do I need dedicated hardware for this? I guess my question is: Is it possible to achieve sub-millisecond response times in a "standard" Linux (ie a desktop variant)? Or is it necessary to look into real-time Linux or other specific kernel configuration? As an aside, I'll mention that I've looked into libftdi as an alternative (could it be?), but since it specifically unloads the kernel driver I'm not sure I should be asking about that here :-) To be honest, I'm not sure if kernelnewbies is the right place to ask, if not, I'll be happy to take my questions elsewhere :-) All the best, -- Thomas

3 6

gpio: userspace tools
by Subhashini Rao Beerisetty Sept. 3, 2018

Sept. 3, 2018

Hi All, I’m using v4.4 kernel and I see that gpiochipN are created under /sys/class/gpio. Latest kernel has three tools lsgpio, gpio-hammer and gpio-event-mon, all these tools opens the device node under /dev/gpiochipN. Does v4.4 kernel has any similar tools to control GPIO pins by using /sys/class/gpio/gpiochipN? If not, can someone point me how the GPIO pins were controlled in the older kernel(which supports /sys/class/gpio)? Thanks

2 1

Queries on bottom halves
by Abhinav Misra Sept. 3, 2018

Sept. 3, 2018

Hi, I have few queries regarding bottom halves- 1. Interrupts use their own per cpu stack for their execution.Which stack does the bottom halves use for their execution ? 2. Why can't we sleep in bottom halves ? BR,Abhinav

5 7