February 2012 - kernelnewbies@kernelnewbies.org

does anyone know how to communicate between domUs on xen?
by summerxyt Feb. 21, 2012

Feb. 21, 2012

hi, I'm not very sure about if I can ask questions about xen here. I have tow questions. The first one is that I'm looking for methods about inter-domain communication for xen. I know besides network, VMs on the same physical machine can exchange information by event channel and grant table. But are these two ways working in kernel mode as part of drivers, or can I directly use them just like network socket in user mode? I've tried to google it but only got something I can't fully understand. The second questions is I want to do some extra jobs (like check tcp/ip header and change some bits) to packages in the virtual network of xen. As all domUs' packages are passing the back driver of dom0, I want to put this extra action to network back driver of dom0. But I'm not sure which part of the source code has the function of receiving packages from front driver. I think it may be function xen_netbk_tx_action. I'll be very appreciated if someone knows the answer. Thanks! summerxyt

2 2

Fwd: Sysctl migrate to 2.6.35
by Rahul Kumar Feb. 21, 2012

Feb. 21, 2012

---------- Forwarded message ---------- From: Rahul Kumar Date: Monday, February 20, 2012 Subject: Sysctl migrate to 2.6.35 To: kernelnewbies(a)nl.linux.org hi, I have userspace sysctl calls made to sysctl tables configured on a 2.6.24 kernel. I have migrated the sysctl code to 2.6.35 kernel (changed all the sysctl table data strctures) I'm seeing warning msgs 'deprecated sysctl warning' when i make sysctl() calls from userspace. The same is working on 2.6.24. Does anyone have any idea on this. Also, sysctl handler do_sysctl is located in sysctl.c whereas in 2.6.35 is located in sysctl_binary.c. Do I need to make any additional changes Can anyone please let me know.

1 0

Some questions from a kernel newbie
by Royo Ruan Feb. 20, 2012

Feb. 20, 2012

Hi All, I am a newbie from China. I want to fix some bugs for linux kernel (current version or older versions), but this is my first time to join community, I don't know how to start... Please help me answer the following questions! :-) Q1, Are there some development projects for newbies? Shoud I start development with them? Q2, I have almost two years working experience on linux kernel development, could I directly fix bugs for linux kernel? Q3, Could you guys recommend a kernel version to start fixing bugs? Q4, Where can I find some details about how to fix kernel bug? Thanks, Zhengwang (Royo)

3 2

some questions about container_of and user_struct
by loody Feb. 20, 2012

Feb. 20, 2012

hi all: I add below function in free_uid to get which task try to free user_struct struct task_struct *p; struct cred *c; c = container_of(up,struct cred, user); p = container_of(c,struct task_struct,real_cred); printk("%s user %p, cred->user = %p p->pid = %d\n",__func__,up, c->user,p->pid); but p->pid is always 0. I think it is wrong for me to get cred and task by using container_of. 1. from kernel definition below, is it ok that member is pointer? #define container_of(ptr, type, member) 2. is there already exist macros or functions I can use for #1 above? 3. what is user_struct used for? When and under what circumstances kernel will try to release it. 4. since user_struct is allocated by kmem_cache_zalloc, is there api or tool I can monitor it? Thanks for your help,

2 1

linux hrtimer affinity
by Vimal Feb. 20, 2012

Feb. 20, 2012

Hi all, Is there a way to set the affinity of hrtimer callback, so that it executes on a particular logical CPU? The reason is that I have a hrtimer callback that executes a tasklet. If the timer callback executes on a different CPU than the one it was enqueued in, then the tasklet is scheduled on the same CPU, which makes it difficult to reason serialising locks to per-CPU data structures. Any ideas? Thanks, -- Vimal

2 2

Re: [PATCH 1/2] staging: easycap: Clean comment style in easycap_usb_probe()
by Ezequiel García Feb. 19, 2012

Feb. 19, 2012

Hi, 2012/2/18 Dan Carpenter <dan.carpenter(a)oracle.com>: > The first line of the comment here should be a blank line. Ok. > > /* > * blah blah blah > * blah blah blah > */ > >> + /* A new struct easycap is always allocated when interface 0 is probed. >> + * It is not possible here to free any existing struct easycap. >> + * This should have been done by easycap_delete() when the device was >> + * physically unplugged. >> + * The allocated struct easycap is saved for later usage when >> + * interfaces 1 and 2 are probed. >> + */ > ^ > Could you make this asterisk line up with the others? Of course. If I send more patches about easycap driver, can I base it on my current local branch? I mean, based on all these patches? Thanks, Ezequiel.

1 0

Allocation of inode
by Akshay Nehe Feb. 19, 2012

Feb. 19, 2012

Hello, I am trying to write a module in which I need to allocate new inode for ext4 filesystem. I have read that ext4_new_inode is used by kernel to do it, so Is it possible to allocate new from my module? -- Regards, Akshay Nehe.

1 0

scsi host refcounting
by nil acuna Feb. 19, 2012

Feb. 19, 2012

Hi, I'm hot on the trail of how adding a SCSI LLD in the kernel works, but i seem to have lost my bearings. in the scsi_add_host_with_dma() procedure, it device_add()s shost_gendev which has the side effect of incrementing the kobj refcount of its parent. in my case, its a pci_dev.dev. a few lines after, it calls get_device(shost->shost_gendev.parent); which would _also_ increment the parent refcount. shouldn't it just do a single bump instead of two? see snip below: -- snip -- error = device_add(&shost->shost_gendev); if (error) goto out; ....... scsi_host_set_state(shost, SHOST_RUNNING); get_device(shost->shost_gendev.parent); -- snip -- I'm just ocularly following the codes right now so i might have missed a device_put on the parent somewhere. please enlighten me on this. regards

1 0

Question on registering driver fops
by Ezequiel García Feb. 18, 2012

Feb. 18, 2012

El día 14 de febrero de 2012 19:42, Greg KH <greg(a)kroah.com> escribió: > On Tue, Feb 14, 2012 at 07:05:48PM -0300, Ezequiel García wrote: >> I noticed that after registering a video driver with >> "video_register_device" the "open" function gets called. >> The registration is like: >> >> peasycap->video_device.fops = &v4l2_fops; >> peasycap->video_device.minor = -1; >> peasycap->video_device.release = (void *)(&videodev_release); >> >> video_set_drvdata(&(peasycap->video_device), (void *)peasycap); >> >> First question: who calls it and why does it open the device? > > Userspace probably. > I find hard that userspace is who opens the device in this particular case, because I get the "open" call just by plugging in the usb, right after usb_probe. (unless there is a daemon, or udev opens the device for module insertion or something). I will investigate this further. Thanks, Ezequiel.

3 4

Re: linux hrtimer affinity
by Vikram Narayanan Feb. 18, 2012

Feb. 18, 2012

Hello, On 17-Feb-12 00:58:30, srimugunthan dhandapani wrote: > Hi all > As an experiment, i am trying to use softirq instead of using tasklet > in my driver code. Softirqs are meant for very critical processing. Why don't you reconsider? > I was passing the driver's structure pointer to the tasklet. Tasklet or softirq? > The "open_softirq" takes only the function pointer as argument. > Since we can't pass any argument to the softirq, is there any > workaround that is possible to make the structure pointer accessible > to softirq? Anyways, you would've defined the softirq handler in your driver. right? > I just need to do this for taking measurements. Timestamps? HTH, Vikram

1 0