November 2018 - kernelnewbies@kernelnewbies.org

pwm_bl add framebuffer from fbcon
by Christian N Nov. 20, 2018

Nov. 20, 2018

Hi All First at all I introduce my usecase -My board based on ARM (OMPA4) can get a custom splash screen on booting using fbcon. -By Device Tree I can configure screen backlight with desired level. i'm trying to turn on backlight when splash screen is displayed and not before to avoid "white flashes". I read Documentation/devicetree/bindings/pwm/pwm.txt Documentation/devicetree/bindings/leds/pwm-backlight.txt Documentation/fb/... https://www.kernel.org/doc/html/v4.14/admin-guide/kernel-parameters.html Also I noticed that pwm_bl has pwm_backlight_check_fb operation, but I didnt find a clear way to relate frame buffer used by fbcon with pwm_bl related functions or how to customize by DT or kernel config which frame buffer should be used by pwm_bl to turn on when valid data is ready in fb Any hints will be apreciated. Regards

1 0

Why eeprom driver is read-only ?
by Ranran Nov. 20, 2018

Nov. 20, 2018

Hello, What is the reason that kernel driver of eeprom is configured only as read-only ? Is it because the BIOS is stored there ? Is there a way to make it writable ? Thank you, Ran

4 6

High IO load cause kswapd hang and unresponsive system
by Naruto Nguyen Nov. 18, 2018

Nov. 18, 2018

Hi everyone, Recently when I upgrade my kernel from 3.12.61 to kernel 4.4.74, I find my system sometimes is very slow and unresponsive When my system is under a heavy IO load, I see the node start to respond very slow, the IO wait time keeps very high, and the node starts unresponsive. I try to dump all block task at that time and found some interesting things, look like under a heavy load I/O, the kswap start to hang in jbd2_log_wait_commit. Sep 07 12:25:35 myPC kernel: [11039.126514] kswapd0 D 0000000000000000 0 26 2 0x00000000 Sep 07 12:25:35 myPC kernel: [11039.126517] ffff88003da27b78 ffff88000ce08280 ffff88003cdf0680 ffff88003da28000 Sep 07 12:25:35 myPC kernel: [11039.126519] ffff88001a15c888 ffff88001a15c824 ffff88003da27bb8 ffff88001a15c8a0 Sep 07 12:25:35 myPC kernel: [11039.126521] ffff88003da27b90 ffffffff815e4c85 ffff88001a15c800 0000000000013183 Sep 07 12:25:35 myPC kernel: [11039.126522] Call Trace: Sep 07 12:25:35 myPC kernel: [11039.126539] [<ffffffff815e4c85>] schedule+0x35/0x80 Sep 07 12:25:35 myPC kernel: [11039.126561] [<ffffffffa02faada>] jbd2_log_wait_commit+0x8a/0xf0 [jbd2] Sep 07 12:25:35 myPC kernel: [11039.126621] [<ffffffffa03b28a4>] ext4_evict_inode+0x2d4/0x4b0 [ext4] Sep 07 12:25:35 myPC kernel: [11039.126630] [<ffffffff81215a7a>] evict+0xba/0x190 Sep 07 12:25:35 myPC kernel: [11039.126635] [<ffffffff81215b82>] dispose_list+0x32/0x50 Sep 07 12:25:35 myPC kernel: [11039.126639] [<ffffffff81216e52>] prune_icache_sb+0x42/0x50 Sep 07 12:25:35 myPC kernel: [11039.126644] [<ffffffff811ff8f6>] super_cache_scan+0x136/0x180 Sep 07 12:25:35 myPC kernel: [11039.126653] [<ffffffff81196a1b>] shrink_slab.part.41+0x20b/0x3f0 Sep 07 12:25:35 myPC kernel: [11039.126521] ffff88003da27b90 ffffffff815e4c85 ffff88001a15c800 0000000000013183 And after that more and more processes are hitting STAT D, nfsd is hang to wait or the inode deletion to complete in __wait_on_freeing_inode. Sep 07 12:25:35 myPC kernel: [11039.390218] nfsd D 0000000000000000 0 30419 2 0x00000080 Sep 07 12:25:35 myPC kernel: [11039.390221] ffff88000da47a98 ffff8800277ec380 ffff8800374ec580 ffff88000da48000 Sep 07 12:25:35 myPC kernel: [11039.390223] ffff88000da47ad8 ffffc9000011edc8 0000000000000000 ffff88000e92e2f8 Sep 07 12:25:35 myPC kernel: [11039.390225] ffff88000da47ab0 ffffffff815e4c85 ffffffff81c094c0 ffff88003ff9d8b8 Sep 07 12:25:35 myPC kernel: [11039.390225] Call Trace: Sep 07 12:25:35 myPC kernel: [11039.390231] [<ffffffff815e4c85>] schedule+0x35/0x80 Sep 07 12:25:35 myPC kernel: [11039.390236] [<ffffffff812157fc>] __wait_on_freeing_inode+0x9c/0xc0 Sep 07 12:25:35 myPC kernel: [11039.390241] [<ffffffff81215d83>] find_inode_fast+0x73/0xa0 Sep 07 12:25:35 myPC kernel: [11039.390245] [<ffffffff81216052>] iget_locked+0x72/0x190 Sep 07 12:25:35 myPC kernel: [11039.390284] [<ffffffffa03aeb8c>] ext4_iget+0x2c/0xb10 [ext4] Sep 07 12:25:35 myPC kernel: [11039.390314] [<ffffffffa03c1055>] ext4_nfs_get_inode+0x45/0x80 [ext4] Sep 07 12:25:35 myPC kernel: [11039.390320] [<ffffffff812211c4>] generic_fh_to_dentry+0x24/0x30 Sep 07 12:25:35 myPC kernel: [11039.390326] [<ffffffff81277ede>] exportfs_decode_fh+0x5e/0x350 Sep 07 12:25:35 myPC kernel: [11039.390347] [<ffffffffa047749a>] fh_verify+0x2da/0x5a0 [nfsd] Sep 07 12:25:35 myPC kernel: [11039.390362] [<ffffffffa0480933>] nfsd3_proc_getattr+0x63/0xd0 [nfsd] Sep 07 12:25:35 myPC kernel: [11039.390374] [<ffffffffa0473e23>] nfsd_dispatch+0xc3/0x260 [nfsd] Sep 07 12:25:35 myPC kernel: [11039.390411] [<ffffffffa0322108>] svc_process_common+0x418/0x6a0 [sunrpc] Sep 07 12:25:35 myPC kernel: [11039.390439] [<ffffffffa032248d>] svc_process+0xfd/0x1b0 [sunrpc] Sep 07 12:25:35 myPC kernel: [11039.390454] [<ffffffffa047389a>] nfsd+0xea/0x160 [nfsd] That causes the other nfs client is hang to wait nfsd to continue. Sep 07 12:25:35 myPC kernel: [11039.207987] Call Trace: Sep 07 12:25:35 myPC kernel: [11039.207994] [<ffffffff815e4c85>] schedule+0x35/0x80 Sep 07 12:25:35 myPC kernel: [11039.208008] [<ffffffffa031acce>] rpc_wait_bit_killable+0x1e/0xa0 [sunrpc] Sep 07 12:25:35 myPC kernel: [11039.208055] [<ffffffff815e5098>] __wait_on_bit+0x58/0x90 Sep 07 12:25:35 myPC kernel: [11039.208059] [<ffffffff815e513e>] out_of_line_wait_on_bit+0x6e/0x80 Sep 07 12:25:35 myPC kernel: [11039.208082] [<ffffffffa031bd26>] __rpc_execute+0x146/0x440 [sunrpc] Sep 07 12:25:35 myPC kernel: [11039.208104] [<ffffffffa0313049>] rpc_run_task+0x69/0x80 [sunrpc] Sep 07 12:25:35 myPC kernel: [11039.208126] [<ffffffffa031309f>] rpc_call_sync+0x3f/0xa0 [sunrpc] Sep 07 12:25:35 myPC kernel: [11039.204818] [<ffffffff81200d69>] vfs_fstatat+0x49/0x90 Sep 07 12:25:35 myPC kernel: [11039.204821] [<ffffffff812011ea>] SYSC_newstat+0x1a/0x40 Sep 07 12:25:35 myPC kernel: [11039.204826] [<ffffffff815e876e>] entry_SYSCALL_64_fastpath+0x12/0x6d Sep 07 12:25:35 myPC kernel: [11039.207972] DWARF2 unwinder stuck at entry_SYSCALL_64_fastpath+0x12/0x6d I guess under heavy load, somehow the inode cannot be evicted in kswap, so the nfsd just keeps waiting and cause the whole system ends up in unresponsive state. I found some related issues https://www.systutorials.com/linux-kernels/105429/nfsv4-1-prevent-a-3-way-d… (it’s not the same as it’s nfsv4 but just refer for the same symptom.) My system has very limited resource, only 1 core and 1GB RAM and the issue is very easy to happen in a slow disk. However, I did not see the same issue in the older kernel 3.12.61. Could you please let me know if my analysis makes sense, if jbd2_log_wait_commit hang causes __wait_on_freeing_inode wait ? Is it a known issue in kernel? If any kernel change causes the issue?. And if we have any WA to avoid that like preventing kswapd kicks in (I have a 5G swap partition) or to make my system is not unresponsive like that (of course it is not to add more RAM :) ? Thanks a lot Brs, Naruto

2 3

Kernel development virtualization options
by Laurence Rochfort Nov. 16, 2018

Nov. 16, 2018

Hi all, Thus far, I've been testing my kernel patches in a KVM vm with a full distro installation, but it's pretty slow and cumbersome. Valerie Aurora has a blog post on using User-mode Linux, and others mention QEMU with an overlay filesystem and BusyBox. What do people recommend for quick iterative development, with good GDB integration? Cheers, Laurence.

2 1

a question on mapping from PTE to swap entry
by Larry Nov. 15, 2018

Nov. 15, 2018

Hi all, It's quoted that "When a page is swapped out, Linux uses the correspondingPTE to store enough information to locate the page on disk again". I'm curious when multiple process has the same PTE which points to the samepage, how can kernel differenciate which page from swap space should be swappedin? For example, there are 3 processes, each of them has a virtual page that is mapped to a phyisical page PAGE_A, that means the same PTE is mapped to same PAGE Process 1 runs out of its timeslice Process 2 get scheduled, PAGE_A was from Process 1, so it should be swapped out Process 3 get scheduled, PAGE_A was from Process 2, so it should be swapped out Process 1 now comes back, when he use PTE to look for its PAGE_A, it will get two PAGE_A, one is from itself, and the other from Process 2 How does kernel handle this situation? Thanks

2 1

function stack frames in the kernel
by Carter Cheng Nov. 12, 2018

Nov. 12, 2018

Hello, I am wondering how the compiler divines which stack to use for function calls and placement of locals and arguments when a function call is made inside the kernel since the kernel has multiple call stacks. Are function calls handled manually inside kernel code or is there something special inside the compiler for handling this? Thanks, Carter

3 4

Re: Creating compressed backing_store as swapfile
by Austin S. Hemmelgarn Nov. 8, 2018

Nov. 8, 2018

On 11/5/2018 10:58 AM, Adam Borowski wrote: > On Mon, Nov 05, 2018 at 08:31:46PM +0530, Pintu Agarwal wrote: >> Hi, >> >> I have one requirement: >> I wanted to have a swapfile (64MB to 256MB) on my system. >> But I wanted the data to be compressed and stored on the disk in my swapfile. >> [Similar to zram, but compressed data should be moved to disk, instead of RAM]. >> >> Note: I wanted to optimize RAM space, so performance is not important >> right now for our requirement. >> >> So, what are the options available, to perform this in 4.x kernel version. >> My Kernel: 4.9.x >> Board: any - (arm64 mostly). >> >> As I know, following are the choices: >> 1) ZRAM: But it compresses and store data in RAM itself >> 2) frontswap + zswap : Didn't explore much on this, not sure if this >> is helpful for our case. >> 3) Manually creating swapfile: but how to compress it ? >> 4) Any other options ? > > Loop device on any filesystem that can compress (such as btrfs)? The > performance would suck, though -- besides the indirection of loop, btrfs > compresses in blocks of 128KB while swap wants 4KB writes. Other similar > option is qemu-nbd -- it can use compressed disk images and expose them to a > (local) nbd client. Swap on any type of a networked storage device (NBD, iSCSI, ATAoE, etc) served from the local system is _really_ risky. The moment the local server process for the storage device gets forced out to swap, you deadlock. Performance isn't _too_ bad for the BTRFS case though (I've actually tested this before), just make sure you disable direct I/O mode on the loop device, otherwise you run the risk of data corruption.

3 5

Creating compressed backing_store as swapfile
by Pintu Agarwal Nov. 8, 2018

Nov. 8, 2018

Hi, I have one requirement: I wanted to have a swapfile (64MB to 256MB) on my system. But I wanted the data to be compressed and stored on the disk in my swapfile. [Similar to zram, but compressed data should be moved to disk, instead of RAM]. Note: I wanted to optimize RAM space, so performance is not important right now for our requirement. So, what are the options available, to perform this in 4.x kernel version. My Kernel: 4.9.x Board: any - (arm64 mostly). As I know, following are the choices: 1) ZRAM: But it compresses and store data in RAM itself 2) frontswap + zswap : Didn't explore much on this, not sure if this is helpful for our case. 3) Manually creating swapfile: but how to compress it ? 4) Any other options ? Thanks, Pintu

2 2

Unsupported relocation type: R_X86_64_PLT32
by Aleix Roca Nonell Nov. 7, 2018

Nov. 7, 2018

Hi all! In my attempt to write my first useful kernel patch, I'm tracking down a possibly kernel regression on my laptop's wireless driver. I'm trying to use git bisect to narrow down the problem and I'm having trouble compiling a Linux kernel 4.14 from Linus tree in my Arch Linux. The error I'm getting is "Unsupported relocation type: R_X86_64_PLT32" I have read about gcc enabling -fPIC by default from gcc-6 so I have tried compiling with gcc-{8,7,5,4.9} with and without "-fno-pic -no-pic" by running make CC="gcc-5 -fno-pic -no-pic" with no luck. I was expecting those R_X86_64_PLT32 relocations not to be generated with this options but there is something I have still not understood. I have been able to compile the kernel after applying greg's stable tree patch [1] (and [2,3] for gcc-8) but the kernel refuses to boot. Any help/comments on either the problem or the followed methodology will be appreciated! Thank you! :D [1] - https://lore.kernel.org/patchwork/patch/892780/ [2] - https://lore.kernel.org/patchwork/patch/868613/ [3] - https://lore.kernel.org/patchwork/patch/934722/

2 4

error: /boot/vmlinuz-4.18.16-1-default has invalid signature
by Manuel Quintero Fonseca Nov. 6, 2018

Nov. 6, 2018

I Compiled the stable kernel obtained from kernel.org, when starting the grub the error came out: Loadin Linux 4.18.16-1-default .. error: /boot/vmlinuz-4.18.16-1-default has invalid signature. loading initial ramdisk ... error: the kernel must be loaded first The steps for the process were: $ git clone git: // git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git cd linux-stable $ git checkout -b stable v4.18.16 $ cp / boot / config-`uname -r` * .config $ make #make modules_install install #reboot with that it is supposed to be compiled and installed in the system and I configure the grub the source: https://kernelnewbies.org/KernelBuild openSUSE TW NOTE: I tested it on a virtual machine in the same way, and did not frame the error regards

2 2