Re: 2 new commits to Bering-uClibc
Erich Titl <[email protected]>
| Newsgroups | gmane.linux.leaf.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi Andrew Am 21.08.2015 um 11:21 schrieb Andrew: > 21.08.2015 12:05, Erich Titl пишет: >> Hi >> >> Am 21.08.2015 um 10:53 schrieb Andrew:... >>>> >>>> Why keep that thing, it is redundant and needs code to read it. >>>> >>> Just to reduce moddb size if it'll be needed for somebody. There are >>> just some lines of code for handling it. >> >> Could you specify why someone might need this and where the positive >> effect is if he needs modules.sqfs instead? >> > I said earlier - for custom kernel modules case. Pretty rare case, but OK >> To mee it looks like you would need the fully blown modules.sqfs plus >> a reduced moddb instead of a slightly bigger moddb an _no_ modules.sqfs. > moddb causes an one big trouble on kernel update - it should contain all > important kernel modules. Including tens of NIC drivers and tens of > wireless devices drivers. So, if you have, for ex., Emulex NIC on remote > box (which drver isn't included into moddb) - you'll have a headache on > remote upgrade; you'll need a physical access to server for running > modules autodetection or you should prepare moddb manually - ad module > and all its deps. And there's no good solution for this. Upgrade does this automagically, maybe not in _all_ circumstances but I tested it with a variety of drivers. And then, in reality you only need a driver for storage and a driver for uplink. But yes, I see the complexity of the problem but I don't see how limiting moddb backup can improve this. cheers ET ------------------------------------------------------------------------------ _______________________________________________ leaf-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/leaf-devel
smime.p7s
(application/pkcs7-signature, 1.9 KB) - not displayed