systemd ebuild still checks for CONFIG_HOTPLUG. This only saves (saved) 200 bytes, was forced on with 45f035ab9b8f (committed to torvalds/linux.git 2012-09-06 20:26:16 (GMT)), and does not exist at all as of 40b313608ad4ea655addd2ec6cdd106477ae8e15 (committed to torvalds/linux.git 2013-06-03 21:20:18 (GMT)). This check does not block building; thus, it doesn't really matter if existing ebuilds are patched or if it is simply omitted in future ebuilds.
I don't understand you. I'm using 3.10 and I still have CONFIG_HOTPLUG here.
http://cateee.net/lkddb/web-lkddb/HOTPLUG.html found in init/Kconfig The configuration item CONFIG_HOTPLUG: prompt: type: bool depends on: (none) defined in init/Kconfig found in Linux kernels: 3.7–3.10 Support for hot-pluggable devices found in arch/i386/Kconfig The configuration item CONFIG_HOTPLUG: prompt: Support for hot-pluggable devices type: bool depends on: (none) defined in arch/i386/Kconfig found in Linux kernels: 2.6.0–2.6.3 Support for hot-pluggable devices found in init/Kconfig The configuration item CONFIG_HOTPLUG: prompt: Support for hot-pluggable devices type: bool depends on: CONFIG_EXPERT defined in init/Kconfig found in Linux kernels: 2.6.4–2.6.39, 3.0–3.6 Compare with: http://cateee.net/lkddb/web-lkddb/SMP.html Symmetric multi-processing support found in arch/x86/Kconfig The configuration item CONFIG_SMP: prompt: Symmetric multi-processing support type: bool depends on: (none) defined in arch/x86/Kconfig found in Linux kernels: 2.6.24–2.6.39, 3.0–3.10, *3.10+HEAD* (emphasis mine) Meaning, if I understand correctly, CONFIG_HOTPLUG is in kernels 2.6.4–3.10; after 3.7, it was forced on. In current trees, it does not exist. (Ctrl-F config HOTPLUG) https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/tree/init/Kconfig https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/init/Kconfig https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/tree/init/Kconfig (won't be in next stable by Greg KH) Even in current 3.10, if you check init/Kconfig: (https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/tree/init/Kconfig?h=linux-3.10.y) config HOTPLUG def_bool y make nconfig doesn't even show the option because it's forced on. So yes, you have CONFIG_HOTPLUG in .config, but nowhere else.
Given that systemd supports kernels older than 3.10, I don't think removing the check is a good idea. According to the systemd ebuild, we support any kernel >=2.6.39.
This is a legitimate bug, even if it blocking builds. CONFIG_HOTPLUG is not in 3.11 (causing unnecessary warnings) and has been forced on for quite some time. > From: Greg KH <gregkh@linuxfoundation.org> > > The CONFIG_HOTPLUG variable is tough to turn off, and almost all arches > default to it on. > > If you turn it off, it saves you a big 200 or so bytes, and then starts > to cause all sorts of problems as the code paths if the option is > disabled is never really tested, and memory segments start to get thrown > away that driver authors assume will always be present.
*** Bug 484602 has been marked as a duplicate of this bug. ***
And, how are we supposed to solve this? Because I think we are with the usual problem when checking for kernel config: can we assume the kernel config used for the check is the version that will be used at boot? In that case, this should probably be solved at eclass level: the check should only warn about option being disabled if the option exists (if not, it should simply show that the option no longer exists)
(In reply to Alex Xu from comment #4) > This is a legitimate bug, even if it blocking builds. *not
config DEVTMPFS bool "Maintain a devtmpfs filesystem to mount at /dev" depends on HOTPLUG local CONFIG_CHECK="... ~DEVTMPFS ... ~HOTPLUG" Ergo, the HOTPLUG check is not necessary.
Fixed in -207.
