April 2020 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

What is the difference between make oldconfig and make menuconfig?
by Hao Lee April 11, 2020

April 11, 2020

I know make oldconfig is used to update my old .config to a newer one and it will prompt me questions about the new configuration options. However, make menuconfig can also handle old .config file properly and it doesn't prompt any questions. I don't know what happened when menuconfig processes my old .config file. Does it make decisions for me automatically? Thanks. Regards, Hao Lee

2 1

two kernel modules for the same device address in dts
by Tomek The Messenger April 10, 2020

April 10, 2020

Hi I am trying to design some functionality and I wonder if something like that is supported in dts and linux: &i2c1 { default_device: default_device@75 { compatible = "some-name-never-mind"; reg = <0x75>; }; my_device: my_device@0 { compatible = "my-device-which-do-only-some-minor-part"; reg = <0x75>; }; }; You know what I mean, is it possible to declare in dts two kernel modules for the same device with the same address? I don't want to extend default_device functionality, I want put my functionality in separate kernel module. I am currently in the middle of implementation but I wonder if it has sense. Both kernel modules will have in fact: probe(struct i2c_client *client, ...), I wonder if probe won't fail. At least in user space shell "i2cdetect -r 1" returns UU for address 0x75 when kernel module is loaded. So that is why I suspect probe might fail. Another story is what in case somebody asks for access to device at the same time from two kernel modules. I hope function: i2c_smbus_write_byte_data has some kind of spinlock, mutexes and just wait till bus is free, but who knows.:)

2 1

[Scheduler] CFS - What happens to each task's slice if nr_running * min_granularity > sched_latency?
by Evan T Mesterhazy April 7, 2020

April 7, 2020

Hi everyone ~ I've read a few different references on CFS and have been looking through fair.c at the CFS scheduler code. One thing I'm not completely understanding is what happens when too many processes are running such that nr_running * min_granularity > sched_latency I know that the scheduler will expand the period so that each process can run for at least the min_granularity, *but how does that interact with nice numbers*? Here's the code for expanding the period: /* * The idea is to set a period in which each task runs once. * * When there are too many tasks (sched_nr_latency) we have to stretch * this period because otherwise the slices get too small. * * p = (nr <= nl) ? l : l*nr/nl */ static u64 __sched_period(unsigned long nr_running) { if (unlikely(nr_running > sched_nr_latency)) return nr_running * sysctl_sched_min_granularity; else return sysctl_sched_latency; } Here's the code for calculating an individual process's slice. It looks like the weighting formula is used here regardless of whether the period has been expanded. - If that's the case, doesn't that mean that some processes will still get a slice that's smaller than the min_granularity? /* * We calculate the wall-time slice from the period by taking a part * proportional to the weight. * * s = p*P[w/rw] */ static u64 sched_slice(struct cfs_rq *cfs_rq, struct sched_entity *se) { u64 slice = __sched_period(cfs_rq->nr_running + !se->on_rq); for_each_sched_entity(se) { struct load_weight *load; struct load_weight lw; cfs_rq = cfs_rq_of(se); load = &cfs_rq->load; if (unlikely(!se->on_rq)) { lw = cfs_rq->load; update_load_add(&lw, se->load.weight); load = &lw; } slice = __calc_delta(slice, se->load.weight, load); } return slice; } I ran a test by starting five busy processes with a nice level of -10. Next, I launched ~40 busy processes with a nice level of 0 (all procs were set to use the same CPU). I expected CFS to expand the period and assign each process a slice equal to the min granularity. However, the 5 processes with nice = -10 still used considerably more CPU than the other processes. Is __calc_delta in the function above actually expanding the slice further based on the nice weighting of the tasks? The __calc_delta function is a bit difficult to follow, so I haven't quite figured out what it's doing. tl;dr I know that CFS expands the period if lots of processes are running. What I'm not sure about is how nice levels affect the slice each task gets if the period has been expanded due to a high number of running tasks. Thanks!

3 3

mmap not working in device driver
by s.achterop＠rug.nl April 6, 2020

April 6, 2020

Hello list, I created a simple misc-device platform device driver, including a test program, see https://github.com/SietseAchterop/Batradio/blob/master/batradio_module/test… https://github.com/SietseAchterop/Batradio/blob/master/batradio_module/batr… This on an raspberry zero with the latest kernel from https://github.com/raspberrypi/linux, compiled from source using arm-linux-gnueabihf-gcc (GCC) 8.3.0. The driver only implements the mmap and ioctl functions. An area in the driver (fiq_buf) is made available to the user program via mmap. The problem is that changing data (status value) in the driver via ioctl, is NOT reflected in the user program. Here a few snippets from the driver: From the init function: batradio_data = devm_kzalloc(&pdev->dev, sizeof(*batradio_data), GFP_KERNEL); if (!batradio_data) return -ENOMEM; batradio_data->fiq_base = devm_kzalloc(&pdev->dev, FIQ_BUFFER_SIZE, // 256*1024 GFP_KERNEL); if (!batradio_data->fiq_base) { dev_err(&pdev->dev, "Couldn't allocate memory!\n"); return -ENOMEM; } fiq_buf = (struct fiq_buffer *)batradio_data->fiq_base; fiq_buf->status = 55; And from the mmap function: size_t size = vma->vm_end - vma->vm_start; unsigned long offset = vma->vm_pgoff << PAGE_SHIFT; if (offset + size > FIQ_BUFFER_SIZE) return -EINVAL; offset += virt_to_phys(batradio_data->fiq_base); vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); if (remap_pfn_range(vma, vma->vm_start, offset >> PAGE_SHIFT, size, vma->vm_page_prot)) { return -EAGAIN; } And finally from the client program: fiq_fd = open(FIQ_PATH, O_RDWR); if (!fiq_fd) { ret = errno; perror("Couldn't open batradio device, error %d", errno); exit(-1); } fiq_addr = mmap(NULL, FIQ_BUFFER_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED, fiq_fd, 0); if (fiq_addr == MAP_FAILED) { ret = errno; perror("Couldn't map the fiq buffer"); exit(-2); } fiq_buf = (struct fiq_buffer*)fiq_addr; printf("status %d\n", fiq_buf->status); All very standard I think. Why are changes in fiq_buf->status in the driver not shown in the client program? Thanks in advance, Sietse

1 0

block size vs bvec length
by Michele Sorcinelli April 5, 2020

April 5, 2020

I created a simple block device driver with a logical queue block size of 512 bytes. $ cat /sys/block/myblock/queue/physical_block_size 512 $ cat /sys/block/myblock/queue/logical_block_size 512 I used rq_for_each_segment() to print bvec.bv_len of the segments and it appears to be 4096. Why is it 4096 rather than 512? Also writing a block of 4096 bytes with dd to /dev/myblock will result in a single write request, while writing a block of 512 bytes will result in a read request followed by a write request. Can someone explain this behavior? Thanks, Michele.

2 2

how to check regmap start address
by Tomek The Messenger April 4, 2020

April 4, 2020

Hi Let's see how it looks: https://github.com/u-boot/u-boot/blob/master/include/regmap.h I first time ever see such thing like this: struct regmap_range { ulong start; ulong size; }; struct regmap { enum regmap_endianness_t endianness; int range_count; struct regmap_range ranges[0]; }; How I can get access to variable ranges it if is 0 size? Who develops this one and who reviewed:) Anyway If I have pointer to regmap can I check what is start variable in some other way?

3 4

probe fail, what with devm_kzalloc, kcalloc data?
by Tomek The Messenger April 3, 2020

April 3, 2020

Hi What happens with devm_kzalloc and devm_kcalloc data when probe function returns not 0? Is this memory automatically deallocated? Example below: static struct my_data* my_static_pointer; static int somedriver_do_probe(struct platform_device *pdev) { int ret; struct device *dev = &pdev->dev; dev_info(dev, "probe\n"); my_static_pointer= devm_kzalloc(dev, sizeof(struct my_data), GFP_KERNEL); if (! my_static_pointer ) return -ENOMEM; .... let's say memory was also allocated with devm_kcalloc somewhere in the meantime ... //do sthm: if (ret != 0 ) return -EINVAL; //let's assume there we returned from probe, what with allocated so far memory??? ... }

2 1

checkpatch.pl false positive?
by Michele Sorcinelli April 2, 2020

April 2, 2020

I tried to run the checkpatch.pl script as follows: scripts/checkpatch.pl -f drivers/hwmon/dell-smm-hwmon.c And I noticed this kind of warnings: WARNING: Missing a blank line after declarations #122: FILE: drivers/hwmon/dell-smm-hwmon.c:122: + unsigned int eax; + unsigned int ebx __packed; WARNING: Missing a blank line after declarations #123: FILE: drivers/hwmon/dell-smm-hwmon.c:123: + unsigned int ebx __packed; + unsigned int ecx __packed; The offending code looks like this: struct smm_regs { unsigned int eax; unsigned int ebx __packed; unsigned int ecx __packed; unsigned int edx __packed; unsigned int esi __packed; unsigned int edi __packed; }; It looks like a false positive as it excepts a blank line after every struct field declaration. Should it be reported to the checkpatch.pl maintainers?

2 3

proxy_arp setting on a vlan interface is getting reset
by Nagaprabhanjan Bellari April 1, 2020

April 1, 2020

Hi, I need a small help w.r.t proxy_arp setting on an interface - even if I set /sys/net/conf/ipv4/all/proxy_arp to 1, doing a ifdown and an ifup on a vlan interface resets the proxy_arp setting. Can something be done to keep this from happening? Thanks, -nagp

2 1