Hi, It's been more than two weeks since I submitted my patch and got no feedback on it. I am Cc'ing some people who may be related to this.
To sum up : since kernel commit bcd68a70cb0eee556d86d93133aa150319bd9f53 by Geert Uytterhoeven (see http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=bcd68a70cb0eee556d86d93133aa150319bd9f53 ) hwclock can't write the date to the RTC on non-interrupt based computers (like newworld PPCs) which have a wrong date set (like after their battery is dead). This is because to synchronize to a full second, hwclock actually reads the date from the RTC and fails as, since the change to rtc-generic, the kernel returns -1 on a wrong date. As I pointed out earlier, it seems that a lot of RTC drivers explicitly avoid returning -1 because of this bug : they should be fixed some time in the future, but hwclock should be fixed first. My Powerbook, as a lot of others Mac, used to recover from their wrong date easily before this switch (which also caused #535354). Now, this is only possible by manual intervention, like I said earlier in this bug report. On others architectures with interrupt based RTCs, this works OK as the synchronization loop is not triggered and hwclock handle this case properly. This patch only add another "source" to this case (erroring on the synchronization loop). I tested it on my machine and it works OK. I may also send it to debian-powe...@lists.debian.org if you want more feedback. But please, it will be very usefull for powerpc users if this patch is applied. Thanks, benjamin -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
