* Ingo Molnar <mingo@xxxxxxx> wrote: > here's that commit again, standalone, with the crash in > af9005_usb_module_init() embedded. Unfortunately i dont have the .config > anymore and the crash log is half a year old. I have re-enabled the > driver in tip/master so if the crash still triggers i should see it > within 1-2 days and i can do a fuller bugreport. no crashes or build failures so far, so it's looking good! There's one new compiler warning that i had to fix - but that's all. Find the patch attached below. Btw., there's a handful of warning fixes i carry in tip/warnings/* topic branches - you can find them in tip/master via: http://people.redhat.com/mingo/tip.git/README the ones affecting drivers/media/ are: f3e67e2: fix warning in drivers/media/video/usbvision/usbvision-i2c.c 491af31: fix warning in drivers/media/video/cx18/cx18-mailbox.c 728342a: fix warnings in drivers/media/dvb/ttusb-dec/ttusb_dec.c 4dda565: fix warning in drivers/media/dvb/frontends/drx397xD.c a2e4f4a: fix warning in drivers/media/common/tuners/mxl5007t.c 8317adf: fix warning in drivers/media/dvb/dvb-usb/anysee.c this one appears to be GCC warning about a real bug: 8317adf: fix warning in drivers/media/dvb/dvb-usb/anysee.c the others are fixes for false positives - i reviewed the affected code paths and each commit log includes the analysis about why it is a false positive. You might ask, how did these drivers/media/ commits get into tip/master? It is a side-effect of us keeping tip/master continuously -Werror clean for 32-bit and 64-bit x86 on allyes/allno/allmod and randconfigs as well, Those warning fixes (for the whole kernel - including drivers/media/) go into the tip/warnings/* branches. Please have a look and git-cherry-pick the above commits if they are still relevant to your tree and if you agree with them. Thanks, Ingo ------------------------->Subject: fix warning in drivers/media/dvb/dvb-usb/af9005-fe.cFrom: Ingo Molnar <mingo@xxxxxxx>Date: Sat Dec 27 12:49:36 CET 2008 this warning: drivers/media/dvb/dvb-usb/af9005-fe.c: In function ‘af9005_fe_refresh_state’: drivers/media/dvb/dvb-usb/af9005-fe.c:223: warning: ‘loc_abort_count’ may be used uninitialized in this function triggers because GCC does not recognize the (complex but correct) error flowbetween: - af9005_fe_refresh_state() and af9005_get_post_vit_err_cw_count() Annotate it. Signed-off-by: Ingo Molnar <mingo@xxxxxxx>------ drivers/media/dvb/dvb-usb/af9005-fe.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: linux/drivers/media/dvb/dvb-usb/af9005-fe.c===================================================================--- linux.orig/drivers/media/dvb/dvb-usb/af9005-fe.c+++ linux/drivers/media/dvb/dvb-usb/af9005-fe.c@@ -220,7 +220,7 @@ static int af9005_get_post_vit_ber(struc u16 * abort_count) { u32 loc_cw_count = 0, loc_err_count;- u16 loc_abort_count;+ u16 uninitialized_var(loc_abort_count); int ret; ret = _______________________________________________v4l-dvb-maintainer mailing listv4l-dvb-maintainer@xxxxxxxxxxxxxxx://www.linuxtv.org/cgi-bin/mailman/listinfo/v4l-dvb-maintainer