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

How to make /dev/ttyACM0 (and friends) exclusive?
by Jeffrey Walton Oct. 11, 2020

Oct. 11, 2020

Hi Everyone, I'm trying to trackdown a problem using my dialup modem. I have a program that opens the modem and watches caller id's. It flashes the hook when a telemarketer calls. It works well until... When another program opens the modem then my program starts reading intermittent responses intended for the other program. I cannot seem to open the device in exclusive mode. The current open is (I also tried with O_EXCL): int modem = open(device_path, O_RDWR | O_NOCTTY | O_SYNC); I'm not sure if it because udev configures the device incorrectly or the kernel driver is configured incorrectly. Or maybe it is impossible to do. Does anyone know how I can make the device exclusive? Thanks in advance.

6 12

how to perform i2c transaction with disabled irq?
by Tomek The Messenger April 30, 2020

April 30, 2020

Hi If I don't have disabled irqs I can use API i2c_smbus_write_byte in order to perform i2c transaction. But how to make i2c transaction if I have disabled irq? Is any built-in function which lets such operation, where there is some polling implemented? The background is that I do i2c transaction in restart handler (some i2c transaction to hot swap controller) in order to reset board. And linux before entering restart handler list disable interrupts. There is a function to enable back interrupts but in company in which I am working engineers reviewed it and said "it is not allowed, restart must be performed when interrupts are disabled".

2 2

Call flush_scheduled_work in the context of a work routine
by Bogdan Bogush April 30, 2020

April 30, 2020

The comment to flush_scheduled_work states that calling this function in the context of a work routine will lead to deadlock. I think it means following scenario: static void handler(struct work_struct *work) { flush_scheduled_work(); } static DECLARE_WORK(work, handler); schedule_work(&work); If this is true for system workqueue (system_wq) is it also applied to custom workqueue? I want to wait for completion of the workqueues in another workqueue. For example: static void handler(struct work_struct *work) { struct work_data * data = (struct work_data *)work; flush_scheduled_work(); kfree(data); } wq = create_workqueue("wq_test"); data = kmalloc(sizeof(struct work_data), GFP_KERNEL); INIT_WORK(&data->work, handler); queue_work(wq, &data->work);

1 0

iptables and combining additional rule sources
by Jeffrey Walton April 29, 2020

April 29, 2020

Hi Everyone, We are having trouble with our MediaWiki installation on a low-end VM. The VM is servicing a lot of spam traffic, and it is driving cpu usage up to about 80%. The 404's appear to be more expensive then the 200's. GoDaddy wrote to us and told us they were going to suspend our service if we don't get cpu usage down. I experimented with several Apache and MediaWiki plugins and I have a design I like. The plugin scans the URL, detects the problematic URLs, and sends the ip address to a privileged out-of-proc proxy to update iptables. The proxy is privileged and can update iptables rules. It also maintains a database to remove the host after 45 days. The problem I am having is, adding the new information to the existing iptables rules in /etc/sysconfig/iptables. I want to write my rules to a separate file and then tell /etc/sysconfig/iptables to include it at the correct position. I read the iptables(8), iptables-save(8) and iptables-restore(8) man pages, but I don't see how to combine the different sources. How do I tell iptables to include a second external source at a specific location? # iptables --version iptables v1.4.21 Thanks in advance. ========================= Here is an example of /etc/sysconfig/iptables with the position I want to insert the MediaWiki ban rules. # cat /etc/sysconfig/iptables *nat :PREROUTING ACCEPT [4276:232374] :POSTROUTING ACCEPT [270:136514] :OUTPUT ACCEPT [270:136514] COMMIT *filter :INPUT ACCEPT [0:0] :FORWARD ACCEPT [0:0] :OUTPUT ACCEPT [269:205262] -A INPUT -p icmp -j ACCEPT -A INPUT -i lo -j ACCEPT ### I want to insert rules here ### *include my-mediawiki-rules ### Back to normal rules ### -A INPUT -m state --state RELATED,ESTABLISHED -j ACCEPT -A INPUT -p tcp -m state --state NEW -m tcp --dport 22 -j ACCEPT -A INPUT -p tcp -m state --state NEW -m tcp --dport 80 -j ACCEPT -A INPUT -p tcp -m state --state NEW -m tcp --dport 443 -j ACCEPT ...

5 8

Kernel Project : Implement network interface default labeling for Smack
by Suraj Upadhyay April 29, 2020

April 29, 2020

Hello Casey and List, I saw a kernel project on kernelnewbies.org [1]. Which goes like this :- > Implement network interfce default labeling for Smack. > Today all unlabeled networking traffic is treated with one > "ambient" Smack label. This project consists of taking advantage > of mechanisms already in the labeled networking code to allow > different labels to be assigned to packets based on the itnerface > upon which they appeared. > Difficulty 5. I see that the project has not been done yet, as I cannot find any described implementation, correct my if I go wrong here. I really want this to be my first kernel project. I am a relatively new comer to the community, and I want to contribute in all my capacity. Could you please advise me a starting point for this project ? Like the area of the codebase I need to study before attempting this. Any help would be appreciated. [1] : https://kernelnewbies.org/KernelProjects/SmackInterface Thanks, Best wishes, Suraj Upadhyay

2 1

USB driver
by Sohaib Mhmd April 29, 2020

April 29, 2020

Hi everyone, I made a very simple USB driver <https://gist.github.com/smalinux/3d6a5b000a32b97250b2d0c86b90711b>, but the problem is that the probe and disconnect functions never was called. I tried to google for this problem but unfortunately: 1- I don't use usb-storage <https://stackoverflow.com/a/31071943/5688267>, it's a keyboard! (but anyway the same problem exists with any usb-based device) and all driver on my machine are bluetooth related drive (i think!) (as you see from lsmod | grep usb) 2- I tried to rmmod usbhid <https://stackoverflow.com/a/4204608/5688267> but I get error: ERROR: Module usbhid is builtin. I tried to search for "usbhid" in .config file to remove it but I didn't find anything. What should I do? thanks, smalinux [root@hp kernel_mod]# lsmod | grep usb sma_usb 16384 0 btusb 57344 0 btrtl 24576 1 btusb btbcm 20480 1 btusb btintel 32768 1 btusb bluetooth 634880 34 btrtl,btintel,btbcm,bnep,ath3k,btusb,rfcomm I use : - Linux hp 5.6.3 #2 SMP Fri Apr 10 06:45:27 EET 2020 x86_64 x86_64 x86_64 GNU/Linux - Linux buildroot 5.4.3-gf7688b48a-dirty #1 SMP Thu Jan 1 00:00:00 UTC 1970 x86_64 GNU/Linux

3 4

module_platform_driver vs module_init
by Tomek The Messenger April 28, 2020

April 28, 2020

Hi If I have to parse some properties from dts and none file operations are set for user space what is better to use platform driver or just module_init? If I use module_init(my_init) then in my_init I need to use: 'of_find_compatible_node' in order to find node in device tree and parse properties. For platform driver I have it on entry: pdev->dev->of_node BR Tomek

1 0

passing __user function parameter from kernel_module result in EFAULT
by Lev R. Oshvang . April 25, 2020

April 25, 2020

Hi all, I am trying to do a mount from kernel module in kernel 5.4 on x86_64. It fails with -14, I tried to set_fs(USER_DS) before calling do_mount but it did not help. do_mount() returns -EFAUTL when it checks filename validity just at the very first lines of do_mount (). Are there other tricks to do in order to overcome and access_ok checks? Thanks, Lev

2 1

RE: Page Fault
by 양원혁 April 24, 2020

April 24, 2020

Yes, you can use the eax instead of the rax

1 0

Invoke binary from kernel space
by hemanth nandish April 23, 2020

April 23, 2020

Hello, I came across the function: call_usermodehelper This can be used to invoke exec a binary from the kernel space. It ultimately calls: call_usermodehelper_exec_async Which internally does: retval = do_execve <https://elixir.bootlin.com/linux/v4.7/ident/do_execve>( getname_kernel <https://elixir.bootlin.com/linux/v4.7/ident/getname_kernel>( sub_info->path <https://elixir.bootlin.com/linux/v4.7/ident/path>),(const char __user <https://elixir.bootlin.com/linux/v4.7/ident/__user> *const __user <https://elixir.bootlin.com/linux/v4.7/ident/__user> *)sub_info->argv <https://elixir.bootlin.com/linux/v4.7/ident/argv>,(const char __user <https://elixir.bootlin.com/linux/v4.7/ident/__user> *const __user <https://elixir.bootlin.com/linux/v4.7/ident/__user> *)sub_info->envp <https://elixir.bootlin.com/linux/v4.7/ident/envp>); This function works as expected and a new thread is spawned. Now when I try to directly call do_execve within my module by passing the required arguments, I am getting -EFAULT error. This is happening in this function: count(argv, MAX_ARG_STRINGS) What am I missing here? Thank you Hemanth

2 1