On Fri, 2012-07-27 at 06:06 -0600, Tim Gardner wrote:
> On 07/26/2012 03:59 PM, Andy Walls wrote:
> > On Thu, 2012-07-26 at 10:44 -0600, Tim Gardner wrote:
> >> Cc: Mauro Carvalho Chehab <mchehab@xxxxxxxxxxxxx>
> >> Cc: linux-media@xxxxxxxxxxxxxxx
> >> Signed-off-by: Tim Gardner <tim.gardner@xxxxxxxxxxxxx>
> >> ---
> >> drivers/media/video/cx25840/cx25840-firmware.c | 15 ++++++++++++---
> >> 1 file changed, 12 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/media/video/cx25840/cx25840-firmware.c b/drivers/media/video/cx25840/cx25840-firmware.c
> >> index 8150200..b3169f9 100644
> >> --- a/drivers/media/video/cx25840/cx25840-firmware.c
> >> +++ b/drivers/media/video/cx25840/cx25840-firmware.c
> >> @@ -61,6 +61,10 @@ static void end_fw_load(struct i2c_client *client)
> >> cx25840_write(client, 0x803, 0x03);
> >> }
> >>
> >> +#define CX2388x_FIRMWARE "v4l-cx23885-avcore-01.fw"
> >> +#define CX231xx_FIRMWARE "v4l-cx231xx-avcore-01.fw"
> >> +#define CX25840_FIRMWARE "v4l-cx25840.fw"
> >> +
> >> static const char *get_fw_name(struct i2c_client *client)
> >> {
> >> struct cx25840_state *state = to_state(i2c_get_clientdata(client));
> >> @@ -68,10 +72,10 @@ static const char *get_fw_name(struct i2c_client *client)
> >> if (firmware[0])
> >> return firmware;
> >> if (is_cx2388x(state))
> >> - return "v4l-cx23885-avcore-01.fw";
> >> + return CX2388x_FIRMWARE;
> >> if (is_cx231xx(state))
> >> - return "v4l-cx231xx-avcore-01.fw";
> >> - return "v4l-cx25840.fw";
> >> + return CX231xx_FIRMWARE;
> >> + return CX25840_FIRMWARE;
> >> }
> >>
> >> static int check_fw_load(struct i2c_client *client, int size)
> >> @@ -164,3 +168,8 @@ int cx25840_loadfw(struct i2c_client *client)
> >>
> >> return check_fw_load(client, size);
> >> }
> >> +
> >> +MODULE_FIRMWARE(CX2388x_FIRMWARE);
> >> +MODULE_FIRMWARE(CX231xx_FIRMWARE);
> >> +MODULE_FIRMWARE(CX25840_FIRMWARE);
> >> +
> >
> > How will the firmware attribute in the .module_info section be used?
> >
> > For any one model of TV capture device, the cx25840 only needs one of
> > the above firmware files. The others need not exist.
> >
> > I would not want anything in user or kernel space to refuse to load the
> > module just because all 3 firmware files are not present.
> >
> > Regards,
> > Andy
> >
>
> The MODULE_FIRMWARE macro is purely informational. Declaring the
> firmware files that the driver _might_ use helps me pare down the
> external firmware package to just those files that I know it must have.
Would it help if the MODULE_FIRMWARE text was more descriptive, e.g.:
MODULE_FIRMWARE(CX25840_FIRMWARE " CX25840/1/2/3 chips only");
MODULE_FIRMWARE(CX231xx_FIRMWARE " CX23100/1/2 chips only");
MODULE_FIRMWARE(CX2388x_FIRMWARE " CX23885/7/8 chips only");
Similarly for ivtv:
MODULE_FIRMWARE(CX2341X_FIRM_ENC_FILENAME);
MODULE_FIRMWARE(CX2341X_FIRM_DEC_FILENAME " PVR-350 only");
MODULE_FIRMWARE(IVTV_DECODE_INIT_MPEG_FILENAME " PVR-350 only");
and cx18:
MODULE_FIRMWARE(FWFILE); /* v4l-cx23418-dig.fw */
MODULE_FIRMWARE(CX18_CPU_FIRMWARE);
MODULE_FIRMWARE(CX18_APU_FIRMWARE);
MODULE_FIRMWARE(XC2028_DEFAULT_FIRMWARE " Non-HVR-1600 cards");
MODULE_FIRMWARE(FWFILE " Yuan MPC-718 only");
or would that confuse some userspace tools?
>
> This patch will have no runtime impact.
OK. For the original patch:
Acked-by: Andy Walls <awalls@xxxxxxxxxxxxxxxx>
Regards,
Andy
> Note that there are plenty of uses of the MODULE_FIRMWARE macro within
> the kernel.
>
> rtg
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
