-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 11/25/2011 08:05 PM, Petri Rosenström wrote: > I do think that it is useful information for users to know what device > they are running. 100% agree. Same issue here, especially relevant when there are many wireless devices in a network, imagine someone trying to find the one AP out of many to physically access after some years, which is a quite a common situation... On the other hand, I also agree that creating additional machtypes for each alias is unnecessary pollution, as it's not about treating them differently in the kernel but only about exposing a different string in /proc/cpuinfo to the user. The idea to distinguish the device sub-machtype/alias in runtime rather than using the board= kernel boot parameter still does sound promising, as that would not need lots of 99.9% redundant sysupgrade images eating up space and bandwidth and allows to just have symlinks instead, while still being able to report the device name identical to what the user can read on the case in /proc/cpuinfo. Doing that might not be as easy as it sounds, as it might not always be as easy as just matching on something in the board flash, but apparently also involves things like RAM size as the only criteria available, and sometimes there will be no way around passing it as a kernel parameter... Also creating any other kind of additional code to handle that could possibly make things more clear, but in the end also create more mess around a simple problem which already got a pragmatic, but non-idealistic solution...
Even within OpenWrt, the way to handle machtypes varies greatly from target to target, e.g. most ARM-based archs just use the arch-numbers which are officially assigned by ARM, ar2xxx/atheros got some runtime detection, most other MIPS targets seem to use OpenWrt-maintained machtype enumeration afaik. What I can see is the need for a clear decision by the target maintainers on how to handle this information, which is useful for the user, but of course could be in the wiki. Obviously this depends on the (intended?) audience addressed by OpenWrt download images and the way downloads are offered (a smarter web-download dialog could also solve a part of this problem) as well as on where the /lib/ar71xx.sh or /lib/ramips.sh finds the device name information (that doesn't necessarily need to be /proc/cpuinfo but could be uboot-env or a userland-parsed kernel parameter, ...) Just my thoughts about this... -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQIcBAEBAgAGBQJOz/pPAAoJEDy9cKN/1Et0aCwP/2F+HSf+SYq97FWu+MPiA3/5 Nff2UCTpsDur7J1iQhckreTF+dEsKwI+T8dve/CXh13U8sIuybJehefPqbyU5nqc 5u1+KAMH1arWYuBpcKWkdZb7I1nozaXxVwPOF5iRZBBd53zltgdI937IK8qIsKw6 xkqXYG9v/hHGHyZomyH95dD4mXgSRH/7BgPdHK7mSNEFW7iojEkDG/Gi7rqp04HM pUdbxcH9LeyOhO+Vb9mJ1uArfQF6Uoe6aqdW1xG3kt8e8g0FYnappvHZz3/IrLra DQTfnqGB9C3GciwtUhY5YlybOtGy949UKFEBRAH4WOhLGwa3oxfb4npppgTFPlo/ WasNCBbFzHEOFDWr7D99ghWTJe6iG7bRdO+5Wd1ntbRGMRAm4nFRelbToCnX5WMN gp+RItye53fNRGxfKHm7K7HWIjMgpgolv1yO7HJXYAEnefYOcpppludWI7tqhDKb aldw/JYzwXMeS+UmxKp/4rPO5Zxd0sWlwKPiIZhrconlI6mmCGDG6khr1j27Vllu KGx4iZsGPNFvKgjqhdpg82ep1p8lLDPlNx4b7N+LDsvL6CaNDbXq3yCGJZImyQt9 ke3CQ3D+f0Cxnt7u+mkPQ/qf+tNZlEjueWuw50Hfm4WLgmn1mqeFu4eajZIMmOBF vWg7Z5L4dRe3TyJiadru =lY4B -----END PGP SIGNATURE----- _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
