Date: Tue, 05 Mar 2013 12:01:36 -0800 From: Oleksandr Tymoshenko <gonzo@bluezbox.com> To: Daisuke Aoyama <aoyama@peach.ne.jp> Cc: freebsd-arm@freebsd.org Subject: Re: FreeBSD/armv6z/clang on Raspberry Pi 512MB (with U-Boot + ubldr) Message-ID: <51364F20.500@bluezbox.com> In-Reply-To: <2D2EAE5B346840A49A5D99A9BFD1DA1D@ad.peach.ne.jp> References: <B5F827FF91C94FF2AFEE00194A2BB2C5@ad.peach.ne.jp> <B508111FCE534B2CBA61F4D1EC1078D3@ad.peach.ne.jp> <D3ABE3919EA74D668DB060952B5CD8C0@ad.peach.ne.jp> <2659960079254C38ACD2F1DCBB7A1A19@ad.peach.ne.jp> <E48DEAF481F74C69A1BC7A01F2B8E74A@ad.peach.ne.jp> <D867259F89CF44409C2359527D0263D4@ad.peach.ne.jp> <512FD3E8.4000307@bluezbox.com> <2D2EAE5B346840A49A5D99A9BFD1DA1D@ad.peach.ne.jp>
next in thread | previous in thread | raw e-mail | index | archive | help
On 3/1/2013 12:57 AM, Daisuke Aoyama wrote:
>> I'd like to thank you again for your outstanding work.
>> I've just committed three patches based on your code:
>>
>> - Platform DMA support for SDHCI
>> - DMA engine driver
>> - DMA support for BCM2835 SDHCI driver
>
> Thank you. I checked the code and found some bug.
> I don't check why bus_dma and bus_space is slow.
> But using generic code is good.
>
> ----------------------------------------------------------------------
> o INFO_WAIT_RESP is cleared when reset.
> o dreq is overwritten by next call.
> o should reset DMA when error.
>
> quick fix is like this:
>
>
> --- bcm2835_dma.c (revision 247518)
> +++ bcm2835_dma.c (working copy)
> @@ -199,6 +199,7 @@
> /* Reset control block */
> cb = sc->sc_dma_ch[ch].cb;
> bzero(cb, sizeof(cb));
> + cb->info = INFO_WAIT_RESP; /* XXX */
> }
Fixed
>
> static int
> @@ -409,8 +410,10 @@
> return (-1);
>
> info = sc->sc_dma_ch[ch].cb->info;
> - info &= ~INFO_PERMAP_MASK;
> - info |= (dreq << INFO_PERMAP_SHIFT) & INFO_PERMAP_MASK;
> + if (dreq) {
> + info &= ~INFO_PERMAP_MASK;
> + info |= (dreq << INFO_PERMAP_SHIFT) & INFO_PERMAP_MASK;
> + }
>
> if (dreq)
> info |= INFO_S_DREQ;
> @@ -459,8 +462,10 @@
> return (-1);
>
> info = sc->sc_dma_ch[ch].cb->info;
> - info &= ~INFO_PERMAP_MASK;
> - info |= (dreq << INFO_PERMAP_SHIFT) & INFO_PERMAP_MASK;
> + if (dreq) {
> + info &= ~INFO_PERMAP_MASK;
> + info |= (dreq << INFO_PERMAP_SHIFT) & INFO_PERMAP_MASK;
> + }
These are not really required. With current code if dreq is zero it will
overwrite info bits properly while when patched - bits will be left
untouched
from previous setup call.
>
> if (dreq)
> info |= INFO_D_DREQ;
> @@ -615,6 +693,7 @@
> debug & DEBUG_ERROR_MASK, ch->ch);
> bus_write_4(sc->sc_mem, BCM_DMA_DEBUG(ch->ch),
> debug & DEBUG_ERROR_MASK);
> + bcm_dma_reset(sc->sc_dev, ch->ch);
> }
>
Fixed
> if (cs & CS_INT) {
> ----------------------------------------------------------------------
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?51364F20.500>