September 2011 - kernelnewbies@kernelnewbies.org

TCP_DELACK_MIN vs TCP_ATO_MIN
by Daniel Baluta Sept. 22, 2011

Sept. 22, 2011

Hello, RFC2582, Section 4.2 says: "... an ACK SHOULD be generated for at least every second full-sized segment, and MUST be generated within 500 ms of the arrival of the first unacknowledged packet. ". I guess that the delayed ACK timeout is computed in tcp_send_delayed_ack: === void tcp_send_delayed_ack(struct sock *sk) { struct inet_connection_sock *icsk = inet_csk(sk); int ato = icsk->icsk_ack.ato; unsigned long timeout; /* .... */ /* Stay within the limit we were given */ timeout = jiffies + ato; ==== Can one explain what is the difference between TCP_DELACK_MIN and TCP_ATO_MIN, specifically if the timeout (ato) is always in the interval [TCP_DELACK_MIN, TCP_DELACK_MAX] ? I want to make the delayed ack timeout configurable as some guys tried here [1], so for this reason I plan to make TCP_DELACK_MIN and TCP_DELACK_MAX tunable via proc entries. thanks, Daniel. [1] http://kerneltrap.org/mailarchive/linux-netdev/2008/9/9/3245554

2 3

Cannot read using USB Skeleton Driver
by Felix Varghese Sept. 21, 2011

Sept. 21, 2011

Hi, I have been trying to communicate with a custom usb device from a SAM9G20-EK board using the usb-skeleton.c driver example in the linux source. I modified the driver to add my device's vendor and product id. The USB device enumerates with two bulk endpoints - one IN and one OUT. The (modified) skel driver successfully detects my device and I get a minor number allocated. After creating a device file with that minor number, I am able to write into the device by typing 'echo "HI" > /dev/mydevice'. I verified that this data arrives at my device intact. Next, I started sending data back from the device and tried to read it back using 'cat /dev/mydevice'. The problem is that read not only doesn't work, but also makes the app get totally stuck. Ctrl-C doesn't work and I have to reboot the board. Some debugging using printk's led me to the following snippet which seems to be causing the hang. If I change the wait to a wait_for_completion_interruptible, Ctrl-C starts working, but still no data is received. if (!dev->processed_urb) { /* * the URB hasn't been processed * do it now */ wait_for_completion(&dev->bulk_in_completion); dev->bulk_in_copied = 0; dev->processed_urb = 1; } Oh, btw I am using linux kernel 2.6.39.4. Any Ideas anyone?? Regards, Felix.

2 10

load balancing scheduler on linux
by Murali N Sept. 21, 2011

Sept. 21, 2011

Hi, Can somebody give pointers to the load balancing scheduler on Linux. I hope this is the default scheduler which is used on most of the linux systems on SMP environment. I would like to know how the scheduler takes a decision to migrate some of the processes to the another core? I have a scenario like below: If CORE0 is running per say 2 threads which can be handled by CORE0 itself ( by running at higher frequencies ) without enabling the CORE1 ( put into some low power mode ). How the scheduler takes this kind of decision intelligently!!! -- Regards, Murali N

2 3

looking for 3.0 source
by Littlefield, Tyler Sept. 21, 2011

Sept. 21, 2011

Hello: With the recent attacks on kernel.org, it's down and I'm unsure of where to get what I want. I found the 2.6 branch on github, but I'm looking for kernel 3.0's source, not 3.4 or whatever we're on now. Is there somewhere I can get it? -- Take care, Ty Web: http://tds-solutions.net The Aspen project: a light-weight barebones mud engine http://code.google.com/p/aspenmud Sent from my toaster.

6 6

How to deal with two tasks with the same virtual time
by Parmenides Sept. 21, 2011

Sept. 21, 2011

Hi, As far as the CFS scheduler is concerned, every task, whose key is the task's virtual time, will be inserted into a red-black tree to improve the performence of seaching specified task. But, that works well only if any task's vritual time is different with the others tasks'. So, I wonder whether there are chances that two tasks occasionally have the same virtual time. If so, how to deal with it?

1 0

Why do the CFS chase fairness?
by Parmenides Sept. 20, 2011

Sept. 20, 2011

Hi, Current kernel 2.6 have adopted the CFS scheduler which try to ensure that all process can obtain their proportions of allotted processor fairly. I have three questions about it. 1. How to understand unfairness? I think unfairness is caused by processes sharing a processor, but I am not confident of it. According to Love, as far as fairness is concerned, if a processor is a perfectly multitasking processor, all of processes running on it should run simultaneously rather than run in turn. For example, there are two processes which have the same nice value. Both of them may run 10 ms on a perfectly multitasking processor, each at 50% of processor power. This kind of processor is prefect fair. In contrast, if the two processes run in turn on a real processor, each may run 5ms at 100% of processor power. Although in both of the two cases two processes run 10ms totally, but the latter case is unfair because one process runs first, the other has to wait for 5ms. In other words, when processes running on a real processor, some of them has to wait more or less until preceding processes use up their CPU shares. Does unfairness mean that? 2. Why do processes need fairness? Yes, we can argue that now that we human beings need fairness, processes do so. :-) But, are there advantages if the scheduler care about fairness? Just for processes not waiting too long? Or other reasons? 3. What's the drawbacks of traditional schdulers? According to Love, traditional schedulers assign each process a absolut timeslcie which yields a constant switching rate but variable fairness. How to understand 'constant switching rate'? What does cause 'variable fairness'?

3 9

reading devices with input.h
by Volker Poplawski Sept. 20, 2011

Sept. 20, 2011

Hi all, what's the recommended method for processing input device data with linux/input.h (from userspace) ? blocking read()? non-blocking read()? select(), epoll()? Regards ......Volker

1 0

2.6.39 Kernel Changes for FileSystem, get_sb removed
by Abhijit Pawar Sept. 20, 2011

Sept. 20, 2011

Hi list, It seems that the support for get_sb function is removed from kernel 2.6.39 onwards. My code which is working till 2.6.38 series is failing with new kernels. I tried finding out but there isnt anything mentioned. So if this function itself is removed then how kernel is going to find out the superblock and use it? Are we supposed to use the mount function provided? This is what i got from one of the kernel patch email. +[mandatory] + ->get_sb() is gone. Switch to use of ->mount(). Typically it's just +a matter of switching from calling get_sb_... to mount_... and changing the +function type. If you were doing it manually, just switch from setting ->mnt_root +to some pointer to returning that pointer. On errors return ERR_PTR(...). diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt It looks like there are other changes related to Pages are also done. Regards, Abhijit Pawar

3 6

do_gettiimeofday query
by Lal Sept. 20, 2011

Sept. 20, 2011

I am having trouble with do_gettimeofday() function. I have following source code: int my_module_init(void) { struct timeval tv = {0}; do_gettimeofday(&tv); printk("tv.tv_sec = %lu\n", tv.tv_sec); return 0; } module_init(my_module_init); This code is built-in kernel, and therefore my_module_init is called during device_initcall. However the tv_sec value seems to be wrong (I was expecting this value to be absolute number of seconds since 1970), but it prints 11 or 12 (which seems to be number of seconds since last boot). I checked do_gettimeofday function source which gets the value from xtime variable. I could not find who is responsible for initializing xtime. I tried to change module_init by late_initcall_sync. Still it's same. Any help is appreciated. Regards Lal

2 1

profiling functions called in interrupt context
by Amit Nagal Sept. 20, 2011

Sept. 20, 2011

Hi , i want to measure time spend in a callback function called from interrupt context . since the time spend will be in us / ns , what kernel apis can i use to measure it ? also since i want to use it in interrupt context , kernel time measurement apis should not sleep . i am using embedded arm target , and tools like OProfile are not available . can i be guided which kernel time measurement apis shall i use to measure time in a interrupt callback ? Regards Amit Nagal

4 6