March 2018 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

Problem installing kernel due to virtualbox?
by John Whitmore March 14, 2018

March 14, 2018

I've been AWOL working on other things for an age, so just getting back up to speed and building a kernel. Everything built successfully but when I try to install the kernel (64bit laptop) it fails due to a virtualbox error. ERROR (dkms apport): kernel package linux-headers-4.16.0-rc5 is not supported Error! Bad return status for module build on kernel: 4.16.0-rc5 (x86_64) Consult /var/lib/dkms/virtualbox/5.0.40/build/make.log for more information. I've had a look at that log file but it doesn't say a whole lot more: ...linux/scripts/Makefile.build:583: recipe for target '/var/lib/dkms/virtualbox/5.0.40/build/vboxdrv' failed make[5]: *** [/var/lib/dkms/virtualbox/5.0.40/build/vboxdrv] Error 2 Makefile:1552: recipe for target '_module_/var/lib/dkms/virtualbox/5.0.40/build' failed make[4]: *** [_module_/var/lib/dkms/virtualbox/5.0.40/build] Error 2 Makefile:24: recipe for target '__sub-make' failed make[3]: *** [__sub-make] Error 2 An Internet search did throw up a few pages about a virtualbox error [1] but I'm not sure the relevance to installing a Linux Kernel. Unless 'sudo make install' is running an Ubuntu script which tries to mess with virtual box. I have removed virtualbox from this machine and tried again but I've still got the same error message in the install process. Maybe I have to reboot after removing virtualbox but it still seems a bit tightly coupled? [1] https://bugs.launchpad.net/ubuntu/+source/virtualbox/+bug/1729568

1 0

What will happen if I kill a process which is waiting for the retuan ioctl() syscall?
by Douglas Su March 14, 2018

March 14, 2018

If I have these: 1. I have implemented a simple ioctl() system call which just delay a period and return. 2. Call this ioctl() from a userspace process, of course, this process will be blocked until ioctl() return. 3. Before ioctl() return, use `kill -9 <pid>` command terminates the process. Question: 1. Can we instantly kill this process? 2. If we can instantly kill this process, does ioctl() still keep delay after killing? 3. If ioctl() still running even after we terminate this process, where does the return value ioctl() will return to? 4. How to keep the device's state consistent when we kill a process when it is invoking a system call? Best regards. Douglas

5 6

During shutdown/reboot Network Subsystem v/s Storage Subsystem
by jitendra kumar khasdev March 14, 2018

March 14, 2018

Hi All, I am curious to know what exactly happen during shutdown/reboot process of linux kernel and how the network stack and storage stack behaves. eg. when we do shutdown, which (network or storage)stack goes down first? --- Jitendra

2 1

Re: [PATCH] scsi: eata: drop VLA in reorder()
by Linus Torvalds March 13, 2018

March 13, 2018

On Sun, Mar 11, 2018 at 8:08 PM, Tobin C. Harding <tobin(a)apporbit.com> wrote: > > I think we are going to see a recurring theme here. MAX_MAILBOXES==64 > so this patch adds 1536 bytes to the stack on a 64 bit machine or 768 > bytes on a 32 bit machine. Yeah, that's a bit excessive. It probably works, but one or two of those allocations will make the kernel stack really tight, so in general I really would suggest using kmalloc() instead, or figuring out some way to simply shrink the data structures. That said, I wonder if the solution to this particular driver is "delete it". Because the hardware is truly ancient and nobody sane would use it any more. The last patch that seemed to come from an actual _user_ finding a problem was in 2008 (commit 20c09df7eb9c: "[SCSI] eata: fix the data buffer accessors conversion regression"). And even then it apparently took a year for people to have noticed the breakage. But because the person who reported that problem is still around, I'll just add him to the cc, just in case. Arthur Marsh, you have the dubious honor and distinction of being the only person to have apparently used that driver in the last ten years. Do you still have hardware using that? Because maybe it's really time to retire that driver. Linus

5 4

Guidelines for Submitting an RFC PATCH
by Joe Smith March 13, 2018

March 13, 2018

Hi All, I understand the guidelines for submitting a PATCH and they are quite rigorous. What about submitting an RFC, since RFC is just to get early comments do I have to make sure that each patch in the RFC compiles or is it OK if all patches together compile and are there any other corners that I can cut. Regards -- JS

3 8

Re: [PATCH] scsi: eata: drop VLA in reorder()
by Tobin C. Harding March 12, 2018

March 12, 2018

Adding kernel newbies to CC because I pose a few noob questions :) Adding Linus to CC because I quoted him. On Sun, Mar 11, 2018 at 10:06:58PM +0100, Salvatore Mesoraca wrote: > n_ready will always be less than or equal to MAX_MAILBOXES. > So we avoid a VLA[1] and use fixed-length arrays instead. > > [1] https://lkml.org/lkml/2018/3/7/621 > > Signed-off-by: Salvatore Mesoraca <s.mesoraca16(a)gmail.com> > --- > drivers/scsi/eata.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/scsi/eata.c b/drivers/scsi/eata.c > index 6501c33..202cd17 100644 > --- a/drivers/scsi/eata.c > +++ b/drivers/scsi/eata.c > @@ -2096,7 +2096,7 @@ static int reorder(struct hostdata *ha, unsigned long cursec, > unsigned int k, n; > unsigned int rev = 0, s = 1, r = 1; > unsigned int input_only = 1, overlap = 0; > - unsigned long sl[n_ready], pl[n_ready], ll[n_ready]; > + unsigned long sl[MAX_MAILBOXES], pl[MAX_MAILBOXES], ll[MAX_MAILBOXES]; I think we are going to see a recurring theme here. MAX_MAILBOXES==64 so this patch adds 1536 bytes to the stack on a 64 bit machine or 768 bytes on a 32 bit machine. Linus already commented on another VLA removal patch that 768 was a lot of stack space. That comment did, however say 'deep in some transfer call chain'. I don't know what a 'transfer call chain' (the transfer bit) is but is there some heuristic we can use to know how deep is deep? Or more to the point, is there some heuristic we can use to know what is an acceptable amount of stack space to use? As far as this patch is concerned wouldn't a kmalloc (with GFP_ATOMIC) be ok? We are in an interrupt handler, can we assume that since IO has just occurred that the IO will be so slow comparatively that a memory allocation will be quick. (assuming IO since eata.c only requests a single irq line.) thanks, Tobin.

2 1

Re: [PATCH] scsi: eata: drop VLA in reorder()
by valdis.kletnieks＠vt.edu March 12, 2018

March 12, 2018

On Mon, 12 Mar 2018 14:08:34 +1100, "Tobin C. Harding" said: > removal patch that 768 was a lot of stack space. That comment did, > however say 'deep in some transfer call chain'. I don't know what a > 'transfer call chain' (the transfer bit) is but is there some heuristic > we can use to know how deep is deep? Or more to the point, is there some > heuristic we can use to know what is an acceptable amount of stack space > to use? The canonical "why we put kernel stacks on a diet" configuration: Imagine a bunch of ISCSI targets - with IPSec wrapping the connection. Arranged into a software RAID5. With LVM. With encryption on the LV. With XFS on the encrypted LV. And then the in-kernel sharing it out over NFS. With more IPSec wrapping the NFS TCP connection. Oh, and I/O interrupts, just for fun. And most of all of that has to fit their *entire* stack chain into 2 4K pages. Suddenly, that 768 bytes is taking close to 10% of *all* of the stack that all of that call chain has to share. And I see that patch is against scsi/eata.c - which means it can plausibly end up sharing that stack scenario above starting at 'software raid5'. (For bonus points, the alert reader is invited to figure out which stack each of the above actually ends up on. No, it isn't split across enough stacks that taking 768 bytes out of any of them is acceptable.. :)

1 0

Where is the __per_cpu_offset array initialized??
by 落痕 March 12, 2018

March 12, 2018

Hi list, Does anyone can tell me where the array __per_cpu_offset is initialized on x86_64 arch, I remembered it was initialized in function setup_per_cpu_areas, but it seems that it is not the case on x86-64 arch. Thanks, Larry

2 2

Compiling kernel with -Wformat-signedness gcc flag
by Himanshu Jha March 9, 2018

March 9, 2018

Hello everyone, I just compiled the kernel by passing "-Wformat-signedness" flag https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html -Wformat-signedness: If -Wformat is specified, also warn if the format string requires an unsigned argument and the argument is signed and vice versa. Now, the C standard says: "If a conversion specification is invalid, the behavior is undefined. If any argument is not the correct type for the corresponding conversion specification, the behavior is *undefined*." For eg: drivers/iio/trigger/stm32-timer-trigger.c:228:24: warning: format ‘%d’ expects argument of type ‘int’, but argument 3 has type ‘unsigned int’ [-Wformat=] return sprintf(buf, "%d\n", (unsigned int)freq); ~^ ~~~~~~~~~~~~~~~~~~ %d What is the rationale for not including/using this flag in the HOSTCFLAGS in the Makefile ? -- Thanks Himanshu Jha

3 2

QEMU environment setup for Linux Kernel
by Pintu Kumar March 8, 2018

March 8, 2018

Hi, I need some help on QEMU environment setup for ARM so that I am very my custom Linux Kernel changes for arm. I know there are lots of references available on internet. Earlier I spend lots of time and finally I could able to setup basic qemu working but still I was not completely satisfied. No network access, could not copy files, no graphics, multiple shells etc. So, I wanted to hear from other developer how they setup and use qemu environment. I want someone to list step by step guide here. I think it is going to be useful for others as well. Basically, I am looking for the following: 1) Which kernel to choose for arm: versatile, or vexpress ? 2) Where to get the full pre-built rootfs image to use with about kernel ? 3) How to get even the Graphical interface with Qemu ? 4) How to setup networking with Qemu, so that I can transfer/scp file from my host Ubuntu PC ? 5) How to launch multiple terminal with the same Qemu setup ? If anybody have done this before please compile the steps here for easy reference. Thank You! Regards, Pintu

3 2