linux-kernel - Re: earlyprintk=xdbc seems broken

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YcCmjP8GCD+2qQH0@hirez.programming.kicks-ass.net>
Date:   Mon, 20 Dec 2021 16:51:40 +0100
From:   Peter Zijlstra <peterz@...radead.org>
To:     Mathias Nyman <mathias.nyman@...ux.intel.com>
Cc:     Greg KH <gregkh@...uxfoundation.org>,
        Dave Hansen <dave.hansen@...el.com>,
        Lu Baolu <baolu.lu@...ux.intel.com>, x86@...nel.org,
        linux-kernel@...r.kernel.org, linux-usb@...r.kernel.org,
        Chunfeng Yun <chunfeng.yun@...iatek.com>
Subject: Re: earlyprintk=xdbc seems broken

On Mon, Dec 20, 2021 at 04:34:13PM +0200, Mathias Nyman wrote:
> On 17.12.2021 15.55, Peter Zijlstra wrote:
> > On Fri, Dec 17, 2021 at 01:01:43PM +0200, Mathias Nyman wrote:
> >> I can reproduce this.
> >> Looks like problems started when driver converted to readl_poll_timeout_atomic() in:
> >>
> >> 796eed4b2342 usb: early: convert to readl_poll_timeout_atomic()
> > 
> > I can confirm, reverting that solves the boot hang, things aren't quite
> > working for me though.
> > 
> >> Seems to hang when read_poll_timeout_atomic() calls ktime_* functions.
> >> Maybe  it's too early for ktime.
> > 
> > It certainly is, using ktime for delay loops sounds daft to me anyhow.
> > 
> >> After reverting that patch it works again for me.
> > 
> > [    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-5.16.0-rc3+ root=UUID=a652986c-fbc6-4341-85c3-b4ad4402f130 ro debug ignore_loglevel sysrq_always_enabled usbcore.autosuspend=-1 earlyprintk=xdbc force_early_printk sched_verbose ftrace=nop mitigations=off nokaslr
> > ...
> > [    0.000000] xhci_dbc:early_xdbc_parse_parameter: dbgp_num: 0
> > ...
> > [    3.161367] xhci_dbc:early_xdbc_setup_hardware: failed to setup the connection to host
> 
> Ok, this is some other issue. I got the boot messages over USB
> (running minicom at the other end, listening to ttyUSB0)

I have a regular A->A USB3 cable, not the special one with a wire
missing. Given the dbc thing works, I feel this ought to work too.

> > The machine does boot.. but I *am* getting tons of:
> > 
> > [  485.546898] usb usb4-port4: Cannot enable. Maybe the USB cable is bad?
> > [  485.546963] usb usb4-port4: config error
> 
> This is expected when xhci driver takes over after the early dbc driver,
> xhci driver resets xHC controller, and all ports turn to normal host ports again.
> 
> Because of the special cable you now have two hosts connected to each other,
> both trying to enumerate a device.
> 
> This whole transition from earlyprintk xdbc to normal xhci driver is 
> not very userfriendly.

Uhhmm... but but that shouldn't be. What if I want to keep using xdbc as
console after that point? specifically, I'll probably end up having:

 earlyprintk=xdbc,keep

once all this starts working. The whole point of early consoles is that
they're more reliable than regular consoles, not that they're 'early'.

> > However, when I do:
> > 
> > $ echo enable > /sys/bus/pci/devices/0000:00:14.0/dbc
> > 
> 
> Yes, this works as it turns on the DbC feature on in xHC hardware,
> which turns the first USB port into a usb device.

But if this works, why isn't earlyprintk working for me?