netdev - Re: [PATCH v3 net-next 1/2] net: dsa: mv88e6xxx: Add watchdog interrupt handler

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: <877f4z8gto.fsf@weeman.i-did-not-set--mail-host-address--so-tickle-me>
Date:   Thu, 09 Feb 2017 11:33:23 -0500
From:   Vivien Didelot <vivien.didelot@...oirfairelinux.com>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     David Miller <davem@...emloft.net>, netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH v3 net-next 1/2] net: dsa: mv88e6xxx: Add watchdog interrupt handler

Hi Andrew,

Andrew Lunn <andrew@...n.ch> writes:

>> > +static int mv88e6097_watchdog_action(struct mv88e6xxx_chip *chip, int irq)
>> > +{
>> > +	u16 reg;
>> > +
>> > +	mv88e6xxx_g2_read(chip, GLOBAL2_WDOG_CONTROL, &reg);
>> 
>> We should not ignore read errors.
>
> We are in the middle of an interrupt handler. If we get a read error
> here, we are probable one step from a "Kernel Panic -- not syncing:
> attempted to kill idle task".
>
> About the only thing which makes sense is to print a warning
> message. But that really should happen in one central place,
> mv88e6xxx_smi_read(), so it covers all reads everywhere.
>
>> > +	dev_info(chip->dev, "Watchdog event: 0x%04x", reg);
>> > +
>> > +	return IRQ_HANDLED;
>> > +}
>> > +
>> > +static void mv88e6097_watchdog_free(struct mv88e6xxx_chip *chip)
>> > +{
>> > +	u16 reg;
>> > +
>> > +	mv88e6xxx_g2_read(chip, GLOBAL2_WDOG_CONTROL, &reg);
>> > +
>> > +	reg &= ~(GLOBAL2_WDOG_CONTROL_EGRESS_ENABLE |
>> > +		 GLOBAL2_WDOG_CONTROL_QC_ENABLE);
>> > +
>> > +	mv88e6xxx_g2_write(chip, GLOBAL2_WDOG_CONTROL, reg);
>> 
>> Same here.
>
> Again, and do what? We are in the process of unbinding/unloading the
> kernel module. We are going to keep going whatever, and there is no
> mechanism to say an error occurred at this point, other than a printk.
> Again, such a printk should be in mv88e6xxx_smi_write().

I do agree with the fact that there is not much left to do if such error
occurs in that context. But still, functions implemented in global2.c
(or whatever internal SMI device file) should be written regardless when
or how they will be used, in an interrupt handler, in driver probe, etc.

These should be just self-documented basic access functions to Marvell
chip registers. chip.c then implements the interface and glue with DSA,
watchdog class, etc.

Mixing this with other layers logic results in losing the robust and
self documented code that we are trying to implement here.

Thanks,

        Vivien