linux-kernel - Re: [PATCH 2/2] platform/chrome: reserve only the I/O ports required for the MEC EC

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: <Ye95vRREnbaLM+EJ@google.com>
Date:   Tue, 25 Jan 2022 12:17:01 +0800
From:   Tzung-Bi Shih <tzungbi@...gle.com>
To:     "Dustin L. Howett" <dustin@...ett.net>
Cc:     "Dustin L. Howett" <dustin@...ett.net>,
        linux-kernel@...r.kernel.org, Benson Leung <bleung@...omium.org>,
        Guenter Roeck <groeck@...omium.org>
Subject: Re: [PATCH 2/2] platform/chrome: reserve only the I/O ports required
 for the MEC EC

On Tue, Jan 25, 2022 at 11:42:29AM +0800, Tzung-Bi Shih wrote:
> On Tue, Jan 25, 2022 at 9:35 AM Dustin L. Howett <dustin@...ett.net> wrote:
> > diff --git a/drivers/platform/chrome/cros_ec_lpc.c b/drivers/platform/chrome/cros_ec_lpc.c
> > index 458eb59db2ff..06fdfe365710 100644
> > --- a/drivers/platform/chrome/cros_ec_lpc.c
> > +++ b/drivers/platform/chrome/cros_ec_lpc.c
> > @@ -341,9 +341,14 @@ static int cros_ec_lpc_probe(struct platform_device *pdev)
> >         u8 buf[2];
> >         int irq, ret;
> >
> > -       if (!devm_request_region(dev, EC_LPC_ADDR_MEMMAP, EC_MEMMAP_SIZE,
> > -                                dev_name(dev))) {
> > -               dev_err(dev, "couldn't reserve memmap region\n");
> > +       /*
> > +        * The Framework Laptop (and possibly other non-ChromeOS devices)
> > +        * only exposes the eight I/O ports that are required for the Microchip EC.
> > +        * Requesting a larger reservation will fail.
> > +        */
> > +       if (!devm_request_region(dev, EC_HOST_CMD_REGION0,
> > +                                EC_HOST_CMD_MEC_REGION_SIZE, dev_name(dev))) {
> > +               dev_err(dev, "couldn't reserve MEC region\n");
> >                 return -EBUSY;

The original code:
- devm_request_region(dev, EC_LPC_ADDR_MEMMAP, ...) and then
- cros_ec_lpc_ops.read(EC_LPC_ADDR_MEMMAP + EC_MEMMAP_ID, ...).

After the patch:
- devm_request_region(dev, EC_HOST_CMD_REGION0, ...) and then
- cros_ec_lpc_ops.read(EC_LPC_ADDR_MEMMAP + EC_MEMMAP_ID, ...).

Does it work if it reads out of request_region range?

> > @@ -366,17 +377,19 @@ static int cros_ec_lpc_probe(struct platform_device *pdev)
> >                         dev_err(dev, "EC ID not detected\n");
> >                         return -ENODEV;
> >                 }
> > -       }
> >
> > -       if (!devm_request_region(dev, EC_HOST_CMD_REGION0,
> > -                                EC_HOST_CMD_REGION_SIZE, dev_name(dev))) {
> > -               dev_err(dev, "couldn't reserve region0\n");
> > -               return -EBUSY;
> > -       }
> > -       if (!devm_request_region(dev, EC_HOST_CMD_REGION1,
> > -                                EC_HOST_CMD_REGION_SIZE, dev_name(dev))) {
> > -               dev_err(dev, "couldn't reserve region1\n");
> > -               return -EBUSY;
> > +               /* Reserve the remaining I/O ports required by the non-MEC protocol. */
> > +               if (!devm_request_region(dev, EC_HOST_CMD_REGION0 + EC_HOST_CMD_MEC_REGION_SIZE,
> > +                                        EC_HOST_CMD_REGION_SIZE - EC_HOST_CMD_MEC_REGION_SIZE,
> > +                                        dev_name(dev))) {
> > +                       dev_err(dev, "couldn't reserve remainder of region0\n");
> > +                       return -EBUSY;
> > +               }
> > +               if (!devm_request_region(dev, EC_HOST_CMD_REGION1,
> > +                                        EC_HOST_CMD_REGION_SIZE, dev_name(dev))) {
> > +                       dev_err(dev, "couldn't reserve region1\n");
> > +                       return -EBUSY;
> > +               }

The 2 request_region are now guarded by the first "if (buf[0] != 'E' || buf[1] != 'C')".  That is, only non-MEC will request the 2 regions.

Doesn't other MECs (e.g. non-Framework Laptop) need the 2 regions?