Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <0e837a2e859b4c633b068368f3c28ba30fc1af70.camel@mniewoehner.de>
Subject: Re: [linus:master] [tpm, tpm_tis] e644b2f498: RIP:acpi_safe_halt
From:   Michael =?ISO-8859-1?Q?Niew=F6hner?= <linux@mniewoehner.de>
To:     Jerry Snitselaar <jsnitsel@redhat.com>
Cc:     Lino Sanfilippo <l.sanfilippo@kunbus.com>,
        Lukas Wunner <lukas@wunner.de>,
        Jarkko Sakkinen <jarkko@kernel.org>, oe-lkp@lists.linux.dev,
        lkp@intel.com, linux-kernel@vger.kernel.org,
        linux-integrity@vger.kernel.org,
        Philipp Rosenberger <p.rosenberger@kunbus.com>,
        peterz@infradead.org
In-Reply-To: <6hbzssr2byskuiznx6vvatcuvjkrbhscvorzm4qcheh3n6n4zj@2nrfawn6rbst>
References: <202305041325.ae8b0c43-yujie.liu@intel.com>
         <d80b180a569a9f068d3a2614f062cfa3a78af5a6.camel@kernel.org>
         <42ea93a1-3186-b8ff-c317-d51f3e13856e@kunbus.com>
         <20230511141607.GA32208@wunner.de>
         <1a8ecf90-80a4-9aac-95e1-9ce0c4e09ba5@kunbus.com>
         <6nf5n6fdnkhx6taa2mecnsmyw7sdgaz6fbvkqy7jqkyd7jq2u7@ogsi6ije32in>
         <fcdc5a27817b17d91df84bb06ad5d382829d5467.camel@mniewoehner.de>
         <6hbzssr2byskuiznx6vvatcuvjkrbhscvorzm4qcheh3n6n4zj@2nrfawn6rbst>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8BIT
Date:   Tue, 30 May 2023 16:40:26 +0200
MIME-Version: 1.0
User-Agent: Evolution 3.44.4 
Precedence: bulk

On Mon, 2023-05-29 at 13:58 -0700, Jerry Snitselaar wrote:
> On Mon, May 29, 2023 at 05:07:54PM +0200, Michael Niewöhner wrote:
> > Hi Jerry,
> > 
> > On Thu, 2023-05-11 at 07:59 -0700, Jerry Snitselaar wrote:
> > > 
> > > IIRC trying to catch the irq storm didn't work in the L490 case for
> > > some reason, so we might still need the dmi entry for that one.
> > > 
> > > The info that the T490s had a pin wired up wrong came from Lenovo, but
> > > this one even looks to be a different vendor so I wonder how often
> > > this happens or if there is something else going on. Is it possible to
> > > get info about the tpm used in the Inspur system? The datasheet online
> > > doesn't mention it.
> > 
> > Are you sure about T490s? To me the wiring looks right on both s and non-s:
> > Pin
> > 18 / PIRQ# goes to PIRQA# of the PCH/SoC.
> > 
> > However on L490 Pin 18 / PIRQ# is wired wrongly to SERIRQ, which probably is
> > the
> > reason that catching the interrupt storm didn't work: I guess this
> > completely
> > messes up LPC communication and causes way more problems. In this case only
> > a
> > DMI quirk can help.
> > 
> > BR
> > Michael
> > 
> 
> I'm searching to see if I can find the old bug/email where that info
> from Lenovo originated.  It could be that the info was wrong, and
> it was some firmware issue instead. IIRC the the T490s issue could be
> solved with the code looking for the irq storm, but the L490 needed
> the dmi check even with the irq storm checking code.

Tbh I wonder why the T490s suffers from that interrupt storm at all, but that
might be a FW bug (it's not handling the interrupt). L490 definitely needs that
DMI check, right.

> 
> 
> I haven't found the info yet, but I did find some other reports from back
> then.
> 
> Spurious irq reported with 5.5.7, so after the irq reverts in v5.5:
> 
> tpm_tis IFX0785:00: 2.0 TPM
> Hardware name: Entroware Proteus/Proteus, BIOS 1.07.07TE0 11/15/2019

That's actually a Clevo N151CU. According to schematics it's wired correctly to
PIRQA#, so probably a FW bug as well.

> 
> Thinkpad P53
> tpm_tis STM7308:00: 2.0 TPM
> Hardware name: LENOVO 20QNCTO1WW/20QNCTO1WW, BIOS N2NET34W (1.19 ) 11/28/2019
> 
> 
> 
> Reports from before the 5.5 reverts:
> 
> tpm_tis MSFT0101:00: 2.0 TPM
> Hyperbook NH5/Clevo NH55RCQ

PIRQ# wired to GPP_B0 - needs correct setup in firmware -> probably a firmware
bug.

> 
> tpm_tis IFX0785:00: 2.0 TPM 
> Clevo N151CU-derived notebook

Same device as Entroware Proteus.

> 
> 
> Regards,
> Jerry
> 
> > > 
> > > Regards,
> > > Jerry
> > > 
> > > > > Thanks,
> > > > > 
> > > > > Lukas
> > > 
> > 
>