>>>>> "Jochen" == Jochen Friedrich <joc...@scram.de> writes:
Jochen> Hi Russell, >>>> (...) but unfortunately, all the revisions I have tested after >>>> r22295 build okay but have failed to boot successfully. >>>> Currently, as of r23118, I lose the serial port very early in the >>>> boot, immediately after: >>>> >> Bas> From 23118 to 22295 are 823 revisions. So in less than 10 builds Bas> you should be able to bisect the revision causing the Bas> failure. You might even recall which earlier builds already are Bas> found to be broken. >> I did. It is the next one 22296. Sorry I didn't make that more >> clear. Jochen> See bug #7874. The problem seems to be a changed handling of Jochen> nvram variables. broadcom-diag expects a special variable Jochen> "nvram_type" which is no longer present since 2.6.34. Jochen> IMHO, the whole diag driver is a hack should be replaced by Jochen> some standard (sysfs?) interface to the ssb bus invariants + Jochen> standard LED, GPIO and button drivers. By judicious application of printk's (note that in order to get the console back for the WGT634U to see them, I had to revert r22663) I managed to figure out what is visible, and now have at least broadcom-diag detecting the board again with this patch: diff --git a/package/broadcom-diag/src/diag.c b/package/broadcom-diag/src/diag.c index 501b835..8862c28 100644 --- a/package/broadcom-diag/src/diag.c +++ b/package/broadcom-diag/src/diag.c @@ -901,7 +901,8 @@ static struct platform_t __init *platform_detect(void) { char *boardnum, *boardtype, *buf; - if (strcmp(getvar("nvram_type"), "cfe") == 0) + + if (strcmp(getvar("nvram_type"), "") == 0 && strcmp(getvar("sdram_ncdl"), "0x00020080") == 0) return &platforms[WGT634U]; /* Look for a model identifier */ -- Russell Senior ``I have nine fingers; you have ten.'' seni...@aracnet.com _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel