From mboxrd@z Thu Jan 1 00:00:00 1970 From: Shradha Shah Subject: Re: [PATCH net v2 1/2] sfc: use 64-bit writes for PIO. Date: Wed, 28 May 2014 10:32:30 +0100 Message-ID: <5385AD2E.3040705@solarflare.com> References: <5385ABE5.9070406@solarflare.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit Cc: To: Return-path: Received: from webmail.solarflare.com ([12.187.104.25]:17218 "EHLO webmail.solarflare.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752066AbaE1Jcd (ORCPT ); Wed, 28 May 2014 05:32:33 -0400 In-Reply-To: <5385ABE5.9070406@solarflare.com> Sender: netdev-owner@vger.kernel.org List-ID: Please ignore this patch. I have sent out the correct patch with a cover letter. My Apologies for duplication Regards, Shradha On 28/05/14 10:27, Shradha Shah wrote: > From: Jon Cooper > > Patch to open-code the memory copy routines. > 32bit writes over the PCI bus causes data corruption. > > Fixes:ee45fd92c739db5b7950163d91dfe5f016af6d24 > > orig-hg-hash: 853f313def1e5e9c733f980f2b4e6330a7d063ef > Signed-off-by: Shradha Shah > --- > drivers/net/ethernet/sfc/tx.c | 24 +++++++++++++++++++----- > 1 file changed, 19 insertions(+), 5 deletions(-) > > diff --git a/drivers/net/ethernet/sfc/tx.c b/drivers/net/ethernet/sfc/tx.c > index fa94753..bd6a4b8 100644 > --- a/drivers/net/ethernet/sfc/tx.c > +++ b/drivers/net/ethernet/sfc/tx.c > @@ -189,6 +189,20 @@ struct efx_short_copy_buffer { > u8 buf[L1_CACHE_BYTES]; > }; > > +/* Copy in explicit 64-bit writes. */ > +static void efx_memcpy_64(void *dest, void *src, size_t len) > +{ > + uint64_t *src64 = src, *dest64 = dest; > + size_t i, l64 = len / 8; > + > + WARN_ON_ONCE(len % 8 != 0); > + WARN_ON_ONCE(((u8 *)dest - (u8 *) 0) % 8 != 0); > + BUILD_BUG_ON(sizeof(uint64_t) != 8); > + > + for(i = 0; i < l64; ++i) > + dest64[i] = src64[i]; > +} > + > /* Copy to PIO, respecting that writes to PIO buffers must be dword aligned. > * Advances piobuf pointer. Leaves additional data in the copy buffer. > */ > @@ -198,7 +212,7 @@ static void efx_memcpy_toio_aligned(struct efx_nic *efx, u8 __iomem **piobuf, > { > int block_len = len & ~(sizeof(copy_buf->buf) - 1); > > - memcpy_toio(*piobuf, data, block_len); > + efx_memcpy_64(*piobuf, data, block_len); > *piobuf += block_len; > len -= block_len; > > @@ -230,7 +244,7 @@ static void efx_memcpy_toio_aligned_cb(struct efx_nic *efx, u8 __iomem **piobuf, > if (copy_buf->used < sizeof(copy_buf->buf)) > return; > > - memcpy_toio(*piobuf, copy_buf->buf, sizeof(copy_buf->buf)); > + efx_memcpy_64(*piobuf, copy_buf->buf, sizeof(copy_buf->buf)); > *piobuf += sizeof(copy_buf->buf); > data += copy_to_buf; > len -= copy_to_buf; > @@ -245,7 +259,7 @@ static void efx_flush_copy_buffer(struct efx_nic *efx, u8 __iomem *piobuf, > { > /* if there's anything in it, write the whole buffer, including junk */ > if (copy_buf->used) > - memcpy_toio(piobuf, copy_buf->buf, sizeof(copy_buf->buf)); > + efx_memcpy_64(piobuf, copy_buf->buf, sizeof(copy_buf->buf)); > } > > /* Traverse skb structure and copy fragments in to PIO buffer. > @@ -304,8 +318,8 @@ efx_enqueue_skb_pio(struct efx_tx_queue *tx_queue, struct sk_buff *skb) > */ > BUILD_BUG_ON(L1_CACHE_BYTES > > SKB_DATA_ALIGN(sizeof(struct skb_shared_info))); > - memcpy_toio(tx_queue->piobuf, skb->data, > - ALIGN(skb->len, L1_CACHE_BYTES)); > + efx_memcpy_64(tx_queue->piobuf, skb->data, > + ALIGN(skb->len, L1_CACHE_BYTES)); > } > > EFX_POPULATE_QWORD_5(buffer->option, > -- Many Thanks, Regards, Shradha Shah