Re: [PATCH net-next 2/2] eth: fbnic: Add devlink dev flash support

[Lee Trager]

On 10/15/24 3:43 AM, Simon Horman wrote:

> On Mon, Oct 14, 2024 at 12:18:36PM +0100, Vadim Fedorenko wrote:
> > On 12/10/2024 03:34, Lee Trager wrote:
> > > fbnic supports updating firmware using a PLDM image signed and distributed
> > > by Meta. PLDM images are written into stored flashed. Flashing does not
> > > interrupt operation.
> > >
> > > On host reboot the newly flashed UEFI driver will be used. To run new
> > > control or cmrt firmware the NIC must be power cycled.
> > >
> > > Signed-off-by: Lee Trager <lee@xxxxxxxxx>
> ...
>
> > > diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_devlink.c b/drivers/net/ethernet/meta/fbnic/fbnic_devlink.c
> ...
>
> > > +/**
> > > + * fbnic_send_component_table - Send PLDM component table to the firmware
> > > + * @context: PLDM FW update structure
> > > + * @component: The component to send
> > > + * @transfer_flag: Flag indication location in component tables
> > > + *
> > > + * Read relevant data from component table and forward it to the firmware.
> > > + * Check response to verify if the firmware indicates that it wishes to
> > > + * proceed with the update.
> > > + *
> > > + * Return: zero on success
> > > + *	    negative error code on failure
> > > + */
> > > +static int fbnic_send_component_table(struct pldmfw *context,
> > > +				      struct pldmfw_component *component,
> > > +				      u8 transfer_flag)
> > > +{
> > > +	struct device *dev = context->dev;
> > > +	u16 id = component->identifier;
> > > +	u8 test_string[80];
> > > +
> > > +	switch (id) {
> > > +	case QSPI_SECTION_CMRT:
> > > +	case QSPI_SECTION_CONTROL_FW:
> > > +	case QSPI_SECTION_OPTION_ROM:
> > > +		break;
> > > +	default:
> > > +		dev_err(dev, "Unknown component ID %u\n", id);
> > > +		return -EINVAL;
> > > +	}
> > > +
> > > +	dev_dbg(dev, "Sending PLDM component table to firmware\n");
> > > +
> > > +	/* Temp placeholder */
> > > +	memcpy(test_string, component->version_string,
> > > +	       min_t(u8, component->version_len, 79));
> > > +	test_string[min_t(u8, component->version_len, 79)] = 0;
> > Looks like this construction can be replaced with strscpy().
> > There were several patchsets in the tree to use strscpy(), let's follow
> > the pattern.
> While looking at these lines, I'm unsure why min_t() is being used
> instead of min() here. As version_len is unsigned and 79 is a positive
> constant, I believe min() should be fine here.

clang complains if I'm not explicit with the type by using min_t()


/home/ltrager/fbnic/src/fbnic_devlink.c:194:3: warning: comparison of 
distinct pointer types ('typeof (component->version_len) *' (aka 
'unsigned char *') and 'typeof (79) *' (aka 'int *')) 
[-Wcompare-distinct-pointer-types]

  194 | min(component->version_len, 79));

      | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

./include/linux/minmax.h:45:19: note: expanded from macro 'min'

   45 | #define min(x, y) __careful_cmp(x, y, <)

      | ^~~~~~~~~~~~~~~~~~~~~~

./include/linux/minmax.h:36:24: note: expanded from macro '__careful_cmp'

   36 | __builtin_choose_expr(__safe_cmp(x, y), \

      | ^~~~~~~~~~~~~~~~

./include/linux/minmax.h:26:4: note: expanded from macro '__safe_cmp'

   26 |                 (__typecheck(x, y) && __no_side_effects(x, y))

      |                  ^~~~~~~~~~~~~~~~~

./include/linux/minmax.h:20:28: note: expanded from macro '__typecheck'

   20 |         (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))

      |                    ~~~~~~~~~~~~~~ ^  ~~~~~~~~~~~~~~

1 warning generated.

> > > +	dev_info(dev, "PLDMFW: Component ID: %u version %s\n",
> > > +		 id, test_string);
> > > +
> > > +	return 0;
> > > +}