Hi folks,
This fixes multicast listen for pcnet32 on at least powerpc and powerpc64
kernels.
The mcast_table is in memory referenced by the card and so it needs
to be accessed in little endian mode.
Ciao, Marcus
--- linux-2.4.19/drivers/net/pcnet32.c.be 2003-02-05 07:59:27.000000000 +0100
+++ linux-2.4.19/drivers/net/pcnet32.c 2003-02-05 08:00:22.000000000 +0100
@@ -1534,7 +1534,9 @@
crc = ether_crc_le(6, addrs);
crc = crc >> 26;
- mcast_table [crc >> 4] |= 1 << (crc & 0xf);
+ mcast_table [crc >> 4] = le16_to_cpu(
+ le16_to_cpu(mcast_table [crc >> 4]) | (1 << (crc & 0xf))
+ );
}
return;
}
On Wed, 5 Feb 2003, Marcus Meissner wrote:
> This fixes multicast listen for pcnet32 on at least powerpc and powerpc64
> kernels.
>
> The mcast_table is in memory referenced by the card and so it needs
> to be accessed in little endian mode.
>
> Ciao, Marcus
>
> --- linux-2.4.19/drivers/net/pcnet32.c.be 2003-02-05 07:59:27.000000000 +0100
> +++ linux-2.4.19/drivers/net/pcnet32.c 2003-02-05 08:00:22.000000000 +0100
> @@ -1534,7 +1534,9 @@
>
> crc = ether_crc_le(6, addrs);
> crc = crc >> 26;
> - mcast_table [crc >> 4] |= 1 << (crc & 0xf);
> + mcast_table [crc >> 4] = le16_to_cpu(
^^^^^^^^^^^
> + le16_to_cpu(mcast_table [crc >> 4]) | (1 << (crc & 0xf))
> + );
Shouldn't the first conversion be `cpu_to_le16'?
Gr{oetje,eeting}s,
Geert (reading bk-commits)
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
On Sun, 23 Feb 2003, Geert Uytterhoeven wrote:
> On Wed, 5 Feb 2003, Marcus Meissner wrote:
> > This fixes multicast listen for pcnet32 on at least powerpc and powerpc64
> > kernels.
> >
> > The mcast_table is in memory referenced by the card and so it needs
> > to be accessed in little endian mode.
> >
> > Ciao, Marcus
> >
> > --- linux-2.4.19/drivers/net/pcnet32.c.be 2003-02-05 07:59:27.000000000 +0100
> > +++ linux-2.4.19/drivers/net/pcnet32.c 2003-02-05 08:00:22.000000000 +0100
> > @@ -1534,7 +1534,9 @@
> >
> > crc = ether_crc_le(6, addrs);
> > crc = crc >> 26;
> > - mcast_table [crc >> 4] |= 1 << (crc & 0xf);
> > + mcast_table [crc >> 4] = le16_to_cpu(
> ^^^^^^^^^^^
> > + le16_to_cpu(mcast_table [crc >> 4]) | (1 << (crc & 0xf))
> > + );
>
> Shouldn't the first conversion be `cpu_to_le16'?
Ugh, a quick grep shows that this driver _always_ uses `le*_to_cpu()' to
convert from CPU to little endian.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Geert Uytterhoeven wrote:
> On Sun, 23 Feb 2003, Geert Uytterhoeven wrote:
>
>>On Wed, 5 Feb 2003, Marcus Meissner wrote:
>>
>>>This fixes multicast listen for pcnet32 on at least powerpc and powerpc64
>>>kernels.
>>>
>>>The mcast_table is in memory referenced by the card and so it needs
>>>to be accessed in little endian mode.
>>>
>>>Ciao, Marcus
>>>
>>>--- linux-2.4.19/drivers/net/pcnet32.c.be 2003-02-05 07:59:27.000000000 +0100
>>>+++ linux-2.4.19/drivers/net/pcnet32.c 2003-02-05 08:00:22.000000000 +0100
>>>@@ -1534,7 +1534,9 @@
>>>
>>> crc = ether_crc_le(6, addrs);
>>> crc = crc >> 26;
>>>- mcast_table [crc >> 4] |= 1 << (crc & 0xf);
>>>+ mcast_table [crc >> 4] = le16_to_cpu(
>>
>> ^^^^^^^^^^^
>>
>>>+ le16_to_cpu(mcast_table [crc >> 4]) | (1 << (crc & 0xf))
>>>+ );
>>
>>Shouldn't the first conversion be `cpu_to_le16'?
>
>
> Ugh, a quick grep shows that this driver _always_ uses `le*_to_cpu()' to
> convert from CPU to little endian.
Cosmetically you are correct, and I prefer it to be changed eventually.
However programatically, it has no effect, because those cpu_to_foo and
foo_to_cpu functions either swap, or they don't. Direction doesn't
matter terribly much :)
Jeff
On Wed, 26 Feb 2003, Jeff Garzik wrote:
> Geert Uytterhoeven wrote:
> > On Sun, 23 Feb 2003, Geert Uytterhoeven wrote:
> >>On Wed, 5 Feb 2003, Marcus Meissner wrote:
> >>>This fixes multicast listen for pcnet32 on at least powerpc and powerpc64
> >>>kernels.
> >>>
> >>>The mcast_table is in memory referenced by the card and so it needs
> >>>to be accessed in little endian mode.
> >>>
> >>>Ciao, Marcus
> >>>
> >>>--- linux-2.4.19/drivers/net/pcnet32.c.be 2003-02-05 07:59:27.000000000 +0100
> >>>+++ linux-2.4.19/drivers/net/pcnet32.c 2003-02-05 08:00:22.000000000 +0100
> >>>@@ -1534,7 +1534,9 @@
> >>>
> >>> crc = ether_crc_le(6, addrs);
> >>> crc = crc >> 26;
> >>>- mcast_table [crc >> 4] |= 1 << (crc & 0xf);
> >>>+ mcast_table [crc >> 4] = le16_to_cpu(
> >>
> >> ^^^^^^^^^^^
> >>
> >>>+ le16_to_cpu(mcast_table [crc >> 4]) | (1 << (crc & 0xf))
> >>>+ );
> >>
> >>Shouldn't the first conversion be `cpu_to_le16'?
> >
> >
> > Ugh, a quick grep shows that this driver _always_ uses `le*_to_cpu()' to
> > convert from CPU to little endian.
>
> Cosmetically you are correct, and I prefer it to be changed eventually.
>
> However programatically, it has no effect, because those cpu_to_foo and
> foo_to_cpu functions either swap, or they don't. Direction doesn't
> matter terribly much :)
That's true.
BTW, you save one swap by changing the code to
mcast_table [crc >> 4] |= cpu_to_le16(1 << (crc & 0xf));
Because the order of bitwise or and swap doesn't matter.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds