Add Linux 4.14.13

[pmenzel]

Tested on keineahnung and sigusr2.

[thomas]

Hi Paul,
further testing of the stock aacraid-driver yielded some deficits in the sys-fs representation. Aparently it doesn't link the disk devices to enclosure slots, a thing that the off-tree microsemi driver does (and the LSI driver also, so i deem it's 'good-practice' in modern HBA drivers). Indeed it doesn't affect the functionality, but it hampers maintenance and client-tools a bit more than just a bit.
So my proposal is to discuss the reinsertion of the off-tree aacraid module in the office (i.e. IRL:).
Thomas

[pmenzel]

Dear Thomas, Am 10.01.2018 um 15:50 schrieb Thomas Kreitler:

further testing of the stock aacraid-driver yielded some deficits in the sys-fs representation. Aparently it doesn't link the disk devices to enclosure slots, a thing that the off-tree microsemi driver does (and the LSI driver also, so i deem it's 'good-practice' in modern HBA drivers). Indeed it doesn't affect the functionality, but it hampers maintenance and client-tools a bit more than just a bit.

Thank you for the feedback. Could you please send the differences to our internal dvteam list or open an issues and mention it there? Even better, report it to the Linux kernel folks.

So my proposal is to discuss the reinsertion of the off-tree aacraid module in the office (i.e. IRL:).

Understood. In my opinion that’s orthogonal to this pull request though as the 4.14.13 series never included the out of tree driver.

[pmenzel]

@donald asked me to select DEVKMEM for /dev/kmem. The default was changed from 4.9 to 4.14, and he needs it for his debugging scripts.

[thomas]

@pmenzel
here are the aacraid parts with the version updated (for reference see also the file linked below, or #561):

https://github.molgen.mpg.de/mariux64/bee-files/blob/2ccc897af9b0dd4455df27b1fc0d671caa898048/linux-4.9.71-196.bee

#  aacraid
#  https://storage.microsemi.com/en-us/downloads/linux_source/linux_source_code/productid=aha-1000-8e&dn=microsemi+adaptec+hba+1000-8e.php
#  http://download.adaptec.com/raid/aac/linux/aacraid-linux-src-1.2.1-55022.tgz
#  http://twiki.molgen.mpg.de/foswiki/Medv/SoftwareRaid
#  CONFIG_SCSI_AACRAID=m
#
build_aacraid_driver_update_and_do_depmod() {
    local DRIVERBUILD=55022

    pushd .

    echo "aacraid driver"

    mkdir -pv ${B}/aacraid_update
    cd ${B}/aacraid_update
    tar -xf /src/mariux/beeroot/downloads/aacraid_source_1.2.1-${DRIVERBUILD}.tgz
    cd aacraid-1.2.1-${DRIVERBUILD}
    chmod -x *
    make -C ${B} M=${B}/aacraid_update/aacraid-1.2.1-${DRIVERBUILD} modules

    mkdir -pv ${D}/lib/modules/${FULLKERNELVERSION}/updates/drivers/scsi/aacraid
    cp -v aacraid.ko ${D}/lib/modules/${FULLKERNELVERSION}/updates/drivers/scsi/aacraid

    cd ${D}/lib/modules/${FULLKERNELVERSION}
    depmod -a -b ${D} ${FULLKERNELVERSION}

    popd
}

and indeed the function call before the zfs thingy

build_aacraid_driver_update_and_do_depmod # this one ...
build_spl_zfs_extra_modules_and_do_depmod

[thomas]

Here is the location info of the missing sys-fs parts
short way, where as the device part is missing:

#ls -la /sys/class/enclosure/7:0:80:0/Disk001
drwxr-xr-x  3 root system    0 Jan 10 13:08 .
drwxr-xr-x 19 root system    0 Jan 10 13:07 ..
-rw-r--r--  1 root system 4096 Jan 11 12:56 active
lrwxrwxrwx  1 root system    0 Jan 10 13:08 device -> ../../../../../../../port-7:1/end_device-7:1/target7:0:65/7:0:65:0
-rw-r--r--  1 root system 4096 Jan 11 12:56 fault
-rw-r--r--  1 root system 4096 Jan 11 12:56 locate
drwxr-xr-x  2 root system    0 Jan 11 12:56 power
-rw-r--r--  1 root system 4096 Jan 11 12:56 power_status
-r--r--r--  1 root system 4096 Jan 11 12:56 slot
-rw-r--r--  1 root system 4096 Jan 11 12:56 status
-r--r--r--  1 root system 4096 Jan 11 12:56 type
-rw-r--r--  1 root system 4096 Jan 11 12:56 uevent

The true location would be:

/sys/devices/pci0000:00/0000:00:03.0/0000:04:00.0/host7/port-7:16/end_device-7:16/target7:0:80/7:0:80:0/enclosure/7:0:80:0/Disk001

[pmenzel]

@donald, @thomas, I updated the bee file, and rebuild the packages. Please test.

[donald]

tested on sigusr2 (it boots, gdm comes up, can log in, meltdown fixed, /dev/kmem tools work...)

[thomas]

Test on mrtorgue :
Loaded 'Adaptec aacraid driver 1.2.1.55022src', thus it had booted and took the off-tree driver.
Disks and ZFS not tested yet.

[donald]

tested on claptrap (nfs server. no problem with browser on nfs client after nfs server reboot this time)