| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAGETcx_g5fdeSibDv8C2S4WpVekMvCQ9srwR3BwCzCU2z3kk-g@mail.gmail.com>
Date: Wed, 13 Mar 2024 18:48:09 -0700
From: Saravana Kannan <saravanak@...gle.com>
To: Geert Uytterhoeven <geert@...ux-m68k.org>
Cc: Rob Herring <robh+dt@...nel.org>, Frank Rowand <frowand.list@...il.com>,
Hervé Codina <herve.codina@...tlin.com>,
Luca Ceresoli <luca.ceresoli@...tlin.com>, kernel-team@...roid.com,
Rob Herring <robh@...nel.org>, devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] of: property: fw_devlink: Fix stupid bug in
remote-endpoint parsing
On Thu, Feb 29, 2024 at 2:11 AM Geert Uytterhoeven <geert@...ux-m68k.org> wrote:
>
> Hi Saravana,
>
> On Sat, Feb 24, 2024 at 6:25 AM Saravana Kannan <saravanak@...glecom> wrote:
> > Introduced a stupid bug in commit 782bfd03c3ae ("of: property: Improve
> > finding the supplier of a remote-endpoint property") due to a last minute
> > incorrect edit of "index !=0" into "!index". This patch fixes it to be
> > "index > 0" to match the comment right next to it.
> >
> > Reported-by: Luca Ceresoli <luca.ceresoli@...tlin.com>
> > Link: https://lore.kernel.org/lkml/20240223171849.10f9901d@booty/
> > Fixes: 782bfd03c3ae ("of: property: Improve finding the supplier of a remote-endpoint property")
> > Signed-off-by: Saravana Kannan <saravanak@...gle.com>
>
> Thanks for your patch!
>
> > --- a/drivers/of/property.c
> > +++ b/drivers/of/property.c
> > @@ -1304,7 +1304,7 @@ static struct device_node *parse_remote_endpoint(struct device_node *np,
> > int index)
> > {
> > /* Return NULL for index > 0 to signify end of remote-endpoints */
> > - if (!index || strcmp(prop_name, "remote-endpoint"))
> > + if (index > 0 || strcmp(prop_name, "remote-endpoint"))
> > return NULL;
> >
> > return of_graph_get_remote_port_parent(np);
> > --
> > 2.44.0.rc0.258.g7320e95886-goog
>
> After this, the "Fixed dependency cycle" messages I reported to be
> gone in [1] are back.
>
> In fact, they are slightly different, and there are now even more of them:
>
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f7000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f6000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f5000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f4000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f3000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f2000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f1000/ports/port@...ndpoint@0
> -platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f0000/ports/port@...ndpoint@0
> -platform feaa0000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f3000/ports/port@...ndpoint@2
> -platform feaa0000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f2000/ports/port@...ndpoint@2
> -platform feaa0000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f1000/ports/port@...ndpoint@2
> -platform feaa0000.csi2: Fixed dependency cycle(s) with
> /soc/video@...f0000/ports/port@...ndpoint@2
> -platform fead0000.hdmi: Fixed dependency cycle(s) with
> /soc/sound@...00000/ports/port@...ndpoint
> -platform feae0000.hdmi: Fixed dependency cycle(s) with
> /soc/sound@...00000/ports/port@...ndpoint
> -platform feb00000.display: Fixed dependency cycle(s) with
> /soc/hdmi@...e0000/ports/port@...ndpoint
> -platform feb00000.display: Fixed dependency cycle(s) with
> /soc/hdmi@...d0000/ports/port@...ndpoint
> -platform hdmi0-out: Fixed dependency cycle(s) with
> /soc/hdmi@...d0000/ports/port@...ndpoint
> -platform hdmi1-out: Fixed dependency cycle(s) with
> /soc/hdmi@...e0000/ports/port@...ndpoint
> -platform vga-encoder: Fixed dependency cycle(s) with /vga/port/endpoint
> -platform vga-encoder: Fixed dependency cycle(s) with
> /soc/display@...00000/ports/port@...ndpoint
> +platform e6ef0000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef0000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef1000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef1000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef2000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef2000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef3000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef3000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef4000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef5000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef6000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef7000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform ec500000.sound: Fixed dependency cycle(s) with /soc/hdmi@...e0000
> +platform ec500000.sound: Fixed dependency cycle(s) with /soc/hdmi@...d0000
> +platform ec500000.sound: Fixed dependency cycle(s) with
> /soc/i2c@...10000/codec@10
> +platform e6ef7000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef6000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef5000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef4000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef3000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef2000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef1000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform e6ef0000.video: Fixed dependency cycle(s) with /soc/csi2@...80000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f7000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f6000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f5000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f4000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f3000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f2000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f1000
> +platform fea80000.csi2: Fixed dependency cycle(s) with /soc/video@...f0000
> +platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/i2c@...d8000/video-receiver@70
> +platform e6ef3000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef2000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef1000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform e6ef0000.video: Fixed dependency cycle(s) with /soc/csi2@...a0000
> +platform feaa0000.csi2: Fixed dependency cycle(s) with /soc/video@...f3000
> +platform feaa0000.csi2: Fixed dependency cycle(s) with /soc/video@...f2000
> +platform feaa0000.csi2: Fixed dependency cycle(s) with /soc/video@...f1000
> +platform feaa0000.csi2: Fixed dependency cycle(s) with /soc/video@...f0000
> +platform feaa0000.csi2: Fixed dependency cycle(s) with
> /soc/i2c@...d8000/video-receiver@70
> +platform ec500000.sound: Fixed dependency cycle(s) with /soc/hdmi@...d0000
> +platform fead0000.hdmi: Fixed dependency cycle(s) with /soc/sound@...00000
> +platform fead0000.hdmi: Fixed dependency cycle(s) with /soc/display@...00000
> +platform ec500000.sound: Fixed dependency cycle(s) with /soc/hdmi@...e0000
> +platform feae0000.hdmi: Fixed dependency cycle(s) with /soc/sound@...00000
> +platform feae0000.hdmi: Fixed dependency cycle(s) with /soc/display@...00000
> +platform feae0000.hdmi: Fixed dependency cycle(s) with /soc/display@...00000
> +platform fead0000.hdmi: Fixed dependency cycle(s) with /soc/display@...00000
> +platform feb00000.display: Fixed dependency cycle(s) with /soc/hdmi@...e0000
> +platform feb00000.display: Fixed dependency cycle(s) with /soc/hdmi@...d0000
> +platform cvbs-in: Fixed dependency cycle(s) with
> /soc/i2c@...d8000/video-receiver@70
> +platform hdmi-in: Fixed dependency cycle(s) with
> /soc/i2c@...d8000/video-receiver@70
> +platform fead0000.hdmi: Fixed dependency cycle(s) with /hdmi0-out
> +platform hdmi0-out: Fixed dependency cycle(s) with /soc/hdmi@...d0000
> +platform feae0000.hdmi: Fixed dependency cycle(s) with /hdmi1-out
> +platform hdmi1-out: Fixed dependency cycle(s) with /soc/hdmi@...e0000
> +platform vga: Fixed dependency cycle(s) with /vga-encoder
> +platform feb00000.display: Fixed dependency cycle(s) with /vga-encoder
> +platform vga-encoder: Fixed dependency cycle(s) with /vga
> +platform vga-encoder: Fixed dependency cycle(s) with /soc/display@...00000
>
> -i2c 2-0010: Fixed dependency cycle(s) with
> /soc/sound@...00000/ports/port@...ndpoint
> +platform ec500000.sound: Fixed dependency cycle(s) with
> /soc/i2c@...10000/codec@10
>
> -i2c 4-0070: Fixed dependency cycle(s) with
> /soc/csi2@...80000/ports/port@...ndpoint
> -i2c 4-0070: Fixed dependency cycle(s) with
> /soc/csi2@...a0000/ports/port@...ndpoint
> -i2c 4-0070: Fixed dependency cycle(s) with /hdmi-in/port/endpoint
> -i2c 4-0070: Fixed dependency cycle(s) with /cvbs-in/port/endpoint
> +platform feaa0000.csi2: Fixed dependency cycle(s) with
> /soc/i2c@...d8000/video-receiver@70
> +platform fea80000.csi2: Fixed dependency cycle(s) with
> /soc/i2c@...d8000/video-receiver@70
> +i2c 4-0070: Fixed dependency cycle(s) with /soc/csi2@...80000
> +i2c 4-0070: Fixed dependency cycle(s) with /soc/csi2@...a0000
>
> I guess all of that is expected?
Hi Geert,
Greg has picked up my "post-init-providers" series in his driver core.
Once you pull that in, you should be able to use the
post-init-providers property to break these cycles. Basically treat it
like any other supplier binding, but use it to mark the link in the
cycle that's not real. For the remote-endpoints case, you need to list
property at the device level. Not the endpoint, port or ports nodes.
Once you add this property, you should see an increase in the number
of device links that are present after all device probing is done.
Also, a bunch of existing device links should get converted from
sync_state_only device links to normal managed device links. Meaning,
the sync_state_only file under those /class/devlink/<device-link-x>/
folder should change from "1" to "0".
If that's what you see and it works, then go ahead and send out a
patch so that the boards you care about have a more
deterministic/stable probe/suspend/resume ordering.
Thanks,
Saravana
>
> [1] https://lore.kernel.org/all/CAMuHMdVon3mdivZQ0O6D4+va0nGBrUQbDp23bEq661QD=4t7+g@mail.gmail.com/
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@...ux-m68k.org
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds
Powered by blists - more mailing lists