> why a change is being made:
There are 2 variants of the android bootdev device. One is EMMC and
other is UFS. We would be not knowing the android boot storage type at
build time. So, we need to know the storage type at run time(bootup
time) by reading the "androidboot.bootdevice=". We need to distinguish
between EMMC and UFS at bootup time and take appropriate action.
> Information about why support is being added for the
androidboot.bootdevice parameter in the Qualcomm UFS host controller driver:
We will get the "androidboot.bootdevice=" by reading cmdline
case 1: if the boot image header version is 3
We get the bootdevice info by reading /proc/cmdline (
__setup("androidboot.bootdevice=", get_android_boot_dev_v3);
Case 2: if the boot image header version is 4
We get the bootdevice info by reading /proc/bootconfig
(android_boot_dev_v4 = xbc_find_value("androidboot.bootdevice", &vnode))
Qualcomm UFS modules(ufs-qcom.c) calls this
get_storage_boot_device()(ufs-cmdline.c) and get the bootdevice info
and take appropriate action. This code is yet to be upstreamed. Please
let me know if this info convincing.
Thanks,
Chetan
On 7/30/2022 1:36 AM, Bart Van Assche wrote:
On 7/29/22 05:54, Chetan Chinnadagudihundi Ravindranath (Consultant)
(QUIC) wrote:
Please suggest any efficient way of making ufs-cmdline part of vmlinux.
A patch description should not only explain what has been changed but
also why a change is being made. Information about why support is
being added for the androidboot.bootdevice parameter in the Qualcomm
UFS host controller driver is missing from the patch description.
Please also address Christoph's comment about this new command-line
parameter not being used anywhere in the upstream kernel tree.
Thanks,
Bart.
