This is a rewrite of the DMA API for SSB devices.
This is needed, because the old (non-existing) "API" made too many bad
assumptions on the API of the host-bus (PCI).
This introduces an almost complete SSB-DMA-API that maps to the lowlevel
bus-API based on the bustype.
Signed-off-by: Michael Buesch <[email protected]>
---
John, please merge for 2.6.27
This is one huge patch, because it's not really possible to split this
without breaking bisect.
Index: wireless-testing/drivers/net/b44.c
===================================================================
--- wireless-testing.orig/drivers/net/b44.c 2008-06-20 02:04:56.000000000 +0200
+++ wireless-testing/drivers/net/b44.c 2008-06-20 02:07:31.000000000 +0200
@@ -145,25 +145,25 @@ B44_STAT_REG_DECLARE
static inline void b44_sync_dma_desc_for_device(struct ssb_device *sdev,
dma_addr_t dma_base,
unsigned long offset,
enum dma_data_direction dir)
{
- dma_sync_single_range_for_device(sdev->dma_dev, dma_base,
- offset & dma_desc_align_mask,
- dma_desc_sync_size, dir);
+ ssb_dma_sync_single_range_for_device(sdev, dma_base,
+ offset & dma_desc_align_mask,
+ dma_desc_sync_size, dir);
}
static inline void b44_sync_dma_desc_for_cpu(struct ssb_device *sdev,
dma_addr_t dma_base,
unsigned long offset,
enum dma_data_direction dir)
{
- dma_sync_single_range_for_cpu(sdev->dma_dev, dma_base,
- offset & dma_desc_align_mask,
- dma_desc_sync_size, dir);
+ ssb_dma_sync_single_range_for_cpu(sdev, dma_base,
+ offset & dma_desc_align_mask,
+ dma_desc_sync_size, dir);
}
static inline unsigned long br32(const struct b44 *bp, unsigned long reg)
{
return ssb_read32(bp->sdev, reg);
}
@@ -610,16 +610,16 @@ static void b44_tx(struct b44 *bp)
for (cons = bp->tx_cons; cons != cur; cons = NEXT_TX(cons)) {
struct ring_info *rp = &bp->tx_buffers[cons];
struct sk_buff *skb = rp->skb;
BUG_ON(skb == NULL);
- dma_unmap_single(bp->sdev->dma_dev,
- rp->mapping,
- skb->len,
- DMA_TO_DEVICE);
+ ssb_dma_unmap_single(bp->sdev,
+ rp->mapping,
+ skb->len,
+ DMA_TO_DEVICE);
rp->skb = NULL;
dev_kfree_skb_irq(skb);
}
bp->tx_cons = cons;
if (netif_queue_stopped(bp->dev) &&
@@ -650,35 +650,35 @@ static int b44_alloc_rx_skb(struct b44 *
dest_idx = dest_idx_unmasked & (B44_RX_RING_SIZE - 1);
map = &bp->rx_buffers[dest_idx];
skb = netdev_alloc_skb(bp->dev, RX_PKT_BUF_SZ);
if (skb == NULL)
return -ENOMEM;
- mapping = dma_map_single(bp->sdev->dma_dev, skb->data,
- RX_PKT_BUF_SZ,
- DMA_FROM_DEVICE);
+ mapping = ssb_dma_map_single(bp->sdev, skb->data,
+ RX_PKT_BUF_SZ,
+ DMA_FROM_DEVICE);
/* Hardware bug work-around, the chip is unable to do PCI DMA
to/from anything above 1GB :-( */
- if (dma_mapping_error(mapping) ||
+ if (ssb_dma_mapping_error(bp->sdev, mapping) ||
mapping + RX_PKT_BUF_SZ > DMA_30BIT_MASK) {
/* Sigh... */
- if (!dma_mapping_error(mapping))
- dma_unmap_single(bp->sdev->dma_dev, mapping,
- RX_PKT_BUF_SZ, DMA_FROM_DEVICE);
+ if (!ssb_dma_mapping_error(bp->sdev, mapping))
+ ssb_dma_unmap_single(bp->sdev, mapping,
+ RX_PKT_BUF_SZ, DMA_FROM_DEVICE);
dev_kfree_skb_any(skb);
skb = __netdev_alloc_skb(bp->dev, RX_PKT_BUF_SZ, GFP_ATOMIC|GFP_DMA);
if (skb == NULL)
return -ENOMEM;
- mapping = dma_map_single(bp->sdev->dma_dev, skb->data,
- RX_PKT_BUF_SZ,
- DMA_FROM_DEVICE);
- if (dma_mapping_error(mapping) ||
+ mapping = ssb_dma_map_single(bp->sdev, skb->data,
+ RX_PKT_BUF_SZ,
+ DMA_FROM_DEVICE);
+ if (ssb_dma_mapping_error(bp->sdev, mapping) ||
mapping + RX_PKT_BUF_SZ > DMA_30BIT_MASK) {
- if (!dma_mapping_error(mapping))
- dma_unmap_single(bp->sdev->dma_dev, mapping, RX_PKT_BUF_SZ,DMA_FROM_DEVICE);
+ if (!ssb_dma_mapping_error(bp->sdev, mapping))
+ ssb_dma_unmap_single(bp->sdev, mapping, RX_PKT_BUF_SZ,DMA_FROM_DEVICE);
dev_kfree_skb_any(skb);
return -ENOMEM;
}
}
rh = (struct rx_header *) skb->data;
@@ -747,15 +747,15 @@ static void b44_recycle_rx(struct b44 *b
if (bp->flags & B44_FLAG_RX_RING_HACK)
b44_sync_dma_desc_for_device(bp->sdev, bp->rx_ring_dma,
dest_idx * sizeof(dest_desc),
DMA_BIDIRECTIONAL);
- dma_sync_single_for_device(bp->sdev->dma_dev, le32_to_cpu(src_desc->addr),
- RX_PKT_BUF_SZ,
- DMA_FROM_DEVICE);
+ ssb_dma_sync_single_for_device(bp->sdev, le32_to_cpu(src_desc->addr),
+ RX_PKT_BUF_SZ,
+ DMA_FROM_DEVICE);
}
static int b44_rx(struct b44 *bp, int budget)
{
int received;
u32 cons, prod;
@@ -769,13 +769,13 @@ static int b44_rx(struct b44 *bp, int bu
struct ring_info *rp = &bp->rx_buffers[cons];
struct sk_buff *skb = rp->skb;
dma_addr_t map = rp->mapping;
struct rx_header *rh;
u16 len;
- dma_sync_single_for_cpu(bp->sdev->dma_dev, map,
+ ssb_dma_sync_single_for_cpu(bp->sdev, map,
RX_PKT_BUF_SZ,
DMA_FROM_DEVICE);
rh = (struct rx_header *) skb->data;
len = le16_to_cpu(rh->len);
if ((len > (RX_PKT_BUF_SZ - RX_PKT_OFFSET)) ||
(rh->flags & cpu_to_le16(RX_FLAG_ERRORS))) {
@@ -803,14 +803,14 @@ static int b44_rx(struct b44 *bp, int bu
if (len > RX_COPY_THRESHOLD) {
int skb_size;
skb_size = b44_alloc_rx_skb(bp, cons, bp->rx_prod);
if (skb_size < 0)
goto drop_it;
- dma_unmap_single(bp->sdev->dma_dev, map,
- skb_size, DMA_FROM_DEVICE);
+ ssb_dma_unmap_single(bp->sdev, map,
+ skb_size, DMA_FROM_DEVICE);
/* Leave out rx_header */
skb_put(skb, len + RX_PKT_OFFSET);
skb_pull(skb, RX_PKT_OFFSET);
} else {
struct sk_buff *copy_skb;
@@ -963,31 +963,31 @@ static int b44_start_xmit(struct sk_buff
netif_stop_queue(dev);
printk(KERN_ERR PFX "%s: BUG! Tx Ring full when queue awake!\n",
dev->name);
goto err_out;
}
- mapping = dma_map_single(bp->sdev->dma_dev, skb->data, len, DMA_TO_DEVICE);
- if (dma_mapping_error(mapping) || mapping + len > DMA_30BIT_MASK) {
+ mapping = ssb_dma_map_single(bp->sdev, skb->data, len, DMA_TO_DEVICE);
+ if (ssb_dma_mapping_error(bp->sdev, mapping) || mapping + len > DMA_30BIT_MASK) {
struct sk_buff *bounce_skb;
/* Chip can't handle DMA to/from >1GB, use bounce buffer */
- if (!dma_mapping_error(mapping))
- dma_unmap_single(bp->sdev->dma_dev, mapping, len,
- DMA_TO_DEVICE);
+ if (!ssb_dma_mapping_error(bp->sdev, mapping))
+ ssb_dma_unmap_single(bp->sdev, mapping, len,
+ DMA_TO_DEVICE);
bounce_skb = __dev_alloc_skb(len, GFP_ATOMIC | GFP_DMA);
if (!bounce_skb)
goto err_out;
- mapping = dma_map_single(bp->sdev->dma_dev, bounce_skb->data,
- len, DMA_TO_DEVICE);
- if (dma_mapping_error(mapping) || mapping + len > DMA_30BIT_MASK) {
- if (!dma_mapping_error(mapping))
- dma_unmap_single(bp->sdev->dma_dev, mapping,
- len, DMA_TO_DEVICE);
+ mapping = ssb_dma_map_single(bp->sdev, bounce_skb->data,
+ len, DMA_TO_DEVICE);
+ if (ssb_dma_mapping_error(bp->sdev, mapping) || mapping + len > DMA_30BIT_MASK) {
+ if (!ssb_dma_mapping_error(bp->sdev, mapping))
+ ssb_dma_unmap_single(bp->sdev, mapping,
+ len, DMA_TO_DEVICE);
dev_kfree_skb_any(bounce_skb);
goto err_out;
}
skb_copy_from_linear_data(skb, skb_put(bounce_skb, len), len);
dev_kfree_skb_any(skb);
@@ -1079,26 +1079,26 @@ static void b44_free_rings(struct b44 *b
for (i = 0; i < B44_RX_RING_SIZE; i++) {
rp = &bp->rx_buffers[i];
if (rp->skb == NULL)
continue;
- dma_unmap_single(bp->sdev->dma_dev, rp->mapping, RX_PKT_BUF_SZ,
- DMA_FROM_DEVICE);
+ ssb_dma_unmap_single(bp->sdev, rp->mapping, RX_PKT_BUF_SZ,
+ DMA_FROM_DEVICE);
dev_kfree_skb_any(rp->skb);
rp->skb = NULL;
}
/* XXX needs changes once NETIF_F_SG is set... */
for (i = 0; i < B44_TX_RING_SIZE; i++) {
rp = &bp->tx_buffers[i];
if (rp->skb == NULL)
continue;
- dma_unmap_single(bp->sdev->dma_dev, rp->mapping, rp->skb->len,
- DMA_TO_DEVICE);
+ ssb_dma_unmap_single(bp->sdev, rp->mapping, rp->skb->len,
+ DMA_TO_DEVICE);
dev_kfree_skb_any(rp->skb);
rp->skb = NULL;
}
}
/* Initialize tx/rx rings for packet processing.
@@ -1114,20 +1114,20 @@ static void b44_init_rings(struct b44 *b
b44_free_rings(bp);
memset(bp->rx_ring, 0, B44_RX_RING_BYTES);
memset(bp->tx_ring, 0, B44_TX_RING_BYTES);
if (bp->flags & B44_FLAG_RX_RING_HACK)
- dma_sync_single_for_device(bp->sdev->dma_dev, bp->rx_ring_dma,
- DMA_TABLE_BYTES,
- DMA_BIDIRECTIONAL);
+ ssb_dma_sync_single_for_device(bp->sdev, bp->rx_ring_dma,
+ DMA_TABLE_BYTES,
+ DMA_BIDIRECTIONAL);
if (bp->flags & B44_FLAG_TX_RING_HACK)
- dma_sync_single_for_device(bp->sdev->dma_dev, bp->tx_ring_dma,
- DMA_TABLE_BYTES,
- DMA_TO_DEVICE);
+ ssb_dma_sync_single_for_device(bp->sdev, bp->tx_ring_dma,
+ DMA_TABLE_BYTES,
+ DMA_TO_DEVICE);
for (i = 0; i < bp->rx_pending; i++) {
if (b44_alloc_rx_skb(bp, -1, i) < 0)
break;
}
}
@@ -1141,31 +1141,33 @@ static void b44_free_consistent(struct b
kfree(bp->rx_buffers);
bp->rx_buffers = NULL;
kfree(bp->tx_buffers);
bp->tx_buffers = NULL;
if (bp->rx_ring) {
if (bp->flags & B44_FLAG_RX_RING_HACK) {
- dma_unmap_single(bp->sdev->dma_dev, bp->rx_ring_dma,
- DMA_TABLE_BYTES,
- DMA_BIDIRECTIONAL);
+ ssb_dma_unmap_single(bp->sdev, bp->rx_ring_dma,
+ DMA_TABLE_BYTES,
+ DMA_BIDIRECTIONAL);
kfree(bp->rx_ring);
} else
- dma_free_coherent(bp->sdev->dma_dev, DMA_TABLE_BYTES,
- bp->rx_ring, bp->rx_ring_dma);
+ ssb_dma_free_consistent(bp->sdev, DMA_TABLE_BYTES,
+ bp->rx_ring, bp->rx_ring_dma,
+ GFP_KERNEL);
bp->rx_ring = NULL;
bp->flags &= ~B44_FLAG_RX_RING_HACK;
}
if (bp->tx_ring) {
if (bp->flags & B44_FLAG_TX_RING_HACK) {
- dma_unmap_single(bp->sdev->dma_dev, bp->tx_ring_dma,
- DMA_TABLE_BYTES,
- DMA_TO_DEVICE);
+ ssb_dma_unmap_single(bp->sdev, bp->tx_ring_dma,
+ DMA_TABLE_BYTES,
+ DMA_TO_DEVICE);
kfree(bp->tx_ring);
} else
- dma_free_coherent(bp->sdev->dma_dev, DMA_TABLE_BYTES,
- bp->tx_ring, bp->tx_ring_dma);
+ ssb_dma_free_consistent(bp->sdev, DMA_TABLE_BYTES,
+ bp->tx_ring, bp->tx_ring_dma,
+ GFP_KERNEL);
bp->tx_ring = NULL;
bp->flags &= ~B44_FLAG_TX_RING_HACK;
}
}
/*
@@ -1184,56 +1186,56 @@ static int b44_alloc_consistent(struct b
size = B44_TX_RING_SIZE * sizeof(struct ring_info);
bp->tx_buffers = kzalloc(size, gfp);
if (!bp->tx_buffers)
goto out_err;
size = DMA_TABLE_BYTES;
- bp->rx_ring = dma_alloc_coherent(bp->sdev->dma_dev, size, &bp->rx_ring_dma, gfp);
+ bp->rx_ring = ssb_dma_alloc_consistent(bp->sdev, size, &bp->rx_ring_dma, gfp);
if (!bp->rx_ring) {
/* Allocation may have failed due to pci_alloc_consistent
insisting on use of GFP_DMA, which is more restrictive
than necessary... */
struct dma_desc *rx_ring;
dma_addr_t rx_ring_dma;
rx_ring = kzalloc(size, gfp);
if (!rx_ring)
goto out_err;
- rx_ring_dma = dma_map_single(bp->sdev->dma_dev, rx_ring,
- DMA_TABLE_BYTES,
- DMA_BIDIRECTIONAL);
+ rx_ring_dma = ssb_dma_map_single(bp->sdev, rx_ring,
+ DMA_TABLE_BYTES,
+ DMA_BIDIRECTIONAL);
- if (dma_mapping_error(rx_ring_dma) ||
+ if (ssb_dma_mapping_error(bp->sdev, rx_ring_dma) ||
rx_ring_dma + size > DMA_30BIT_MASK) {
kfree(rx_ring);
goto out_err;
}
bp->rx_ring = rx_ring;
bp->rx_ring_dma = rx_ring_dma;
bp->flags |= B44_FLAG_RX_RING_HACK;
}
- bp->tx_ring = dma_alloc_coherent(bp->sdev->dma_dev, size, &bp->tx_ring_dma, gfp);
+ bp->tx_ring = ssb_dma_alloc_consistent(bp->sdev, size, &bp->tx_ring_dma, gfp);
if (!bp->tx_ring) {
- /* Allocation may have failed due to dma_alloc_coherent
+ /* Allocation may have failed due to ssb_dma_alloc_consistent
insisting on use of GFP_DMA, which is more restrictive
than necessary... */
struct dma_desc *tx_ring;
dma_addr_t tx_ring_dma;
tx_ring = kzalloc(size, gfp);
if (!tx_ring)
goto out_err;
- tx_ring_dma = dma_map_single(bp->sdev->dma_dev, tx_ring,
+ tx_ring_dma = ssb_dma_map_single(bp->sdev, tx_ring,
DMA_TABLE_BYTES,
DMA_TO_DEVICE);
- if (dma_mapping_error(tx_ring_dma) ||
+ if (ssb_dma_mapping_error(bp->sdev, tx_ring_dma) ||
tx_ring_dma + size > DMA_30BIT_MASK) {
kfree(tx_ring);
goto out_err;
}
bp->tx_ring = tx_ring;
Index: wireless-testing/drivers/net/wireless/b43/dma.c
===================================================================
--- wireless-testing.orig/drivers/net/wireless/b43/dma.c 2008-06-20 02:04:56.000000000 +0200
+++ wireless-testing/drivers/net/wireless/b43/dma.c 2008-06-20 02:07:31.000000000 +0200
@@ -325,51 +325,51 @@ static inline
dma_addr_t map_descbuffer(struct b43_dmaring *ring,
unsigned char *buf, size_t len, int tx)
{
dma_addr_t dmaaddr;
if (tx) {
- dmaaddr = dma_map_single(ring->dev->dev->dma_dev,
- buf, len, DMA_TO_DEVICE);
+ dmaaddr = ssb_dma_map_single(ring->dev->dev,
+ buf, len, DMA_TO_DEVICE);
} else {
- dmaaddr = dma_map_single(ring->dev->dev->dma_dev,
- buf, len, DMA_FROM_DEVICE);
+ dmaaddr = ssb_dma_map_single(ring->dev->dev,
+ buf, len, DMA_FROM_DEVICE);
}
return dmaaddr;
}
static inline
void unmap_descbuffer(struct b43_dmaring *ring,
dma_addr_t addr, size_t len, int tx)
{
if (tx) {
- dma_unmap_single(ring->dev->dev->dma_dev,
- addr, len, DMA_TO_DEVICE);
+ ssb_dma_unmap_single(ring->dev->dev,
+ addr, len, DMA_TO_DEVICE);
} else {
- dma_unmap_single(ring->dev->dev->dma_dev,
- addr, len, DMA_FROM_DEVICE);
+ ssb_dma_unmap_single(ring->dev->dev,
+ addr, len, DMA_FROM_DEVICE);
}
}
static inline
void sync_descbuffer_for_cpu(struct b43_dmaring *ring,
dma_addr_t addr, size_t len)
{
B43_WARN_ON(ring->tx);
- dma_sync_single_for_cpu(ring->dev->dev->dma_dev,
- addr, len, DMA_FROM_DEVICE);
+ ssb_dma_sync_single_for_cpu(ring->dev->dev,
+ addr, len, DMA_FROM_DEVICE);
}
static inline
void sync_descbuffer_for_device(struct b43_dmaring *ring,
dma_addr_t addr, size_t len)
{
B43_WARN_ON(ring->tx);
- dma_sync_single_for_device(ring->dev->dev->dma_dev,
- addr, len, DMA_FROM_DEVICE);
+ ssb_dma_sync_single_for_device(ring->dev->dev,
+ addr, len, DMA_FROM_DEVICE);
}
static inline
void free_descriptor_buffer(struct b43_dmaring *ring,
struct b43_dmadesc_meta *meta)
{
@@ -378,43 +378,48 @@ static inline
meta->skb = NULL;
}
}
static int alloc_ringmemory(struct b43_dmaring *ring)
{
- struct device *dma_dev = ring->dev->dev->dma_dev;
gfp_t flags = GFP_KERNEL;
/* The specs call for 4K buffers for 30- and 32-bit DMA with 4K
* alignment and 8K buffers for 64-bit DMA with 8K alignment. Testing
* has shown that 4K is sufficient for the latter as long as the buffer
* does not cross an 8K boundary.
*
* For unknown reasons - possibly a hardware error - the BCM4311 rev
* 02, which uses 64-bit DMA, needs the ring buffer in very low memory,
* which accounts for the GFP_DMA flag below.
+ *
+ * The flags here must match the flags in free_ringmemory below!
*/
if (ring->type == B43_DMA_64BIT)
flags |= GFP_DMA;
- ring->descbase = dma_alloc_coherent(dma_dev, B43_DMA_RINGMEMSIZE,
- &(ring->dmabase), flags);
+ ring->descbase = ssb_dma_alloc_consistent(ring->dev->dev,
+ B43_DMA_RINGMEMSIZE,
+ &(ring->dmabase), flags);
if (!ring->descbase) {
b43err(ring->dev->wl, "DMA ringmemory allocation failed\n");
return -ENOMEM;
}
memset(ring->descbase, 0, B43_DMA_RINGMEMSIZE);
return 0;
}
static void free_ringmemory(struct b43_dmaring *ring)
{
- struct device *dma_dev = ring->dev->dev->dma_dev;
+ gfp_t flags = GFP_KERNEL;
+
+ if (ring->type == B43_DMA_64BIT)
+ flags |= GFP_DMA;
- dma_free_coherent(dma_dev, B43_DMA_RINGMEMSIZE,
- ring->descbase, ring->dmabase);
+ ssb_dma_free_consistent(ring->dev->dev, B43_DMA_RINGMEMSIZE,
+ ring->descbase, ring->dmabase, flags);
}
/* Reset the RX DMA channel */
static int b43_dmacontroller_rx_reset(struct b43_wldev *dev, u16 mmio_base,
enum b43_dmatype type)
{
@@ -515,13 +520,13 @@ static int b43_dmacontroller_tx_reset(st
/* Check if a DMA mapping address is invalid. */
static bool b43_dma_mapping_error(struct b43_dmaring *ring,
dma_addr_t addr,
size_t buffersize, bool dma_to_device)
{
- if (unlikely(dma_mapping_error(addr)))
+ if (unlikely(ssb_dma_mapping_error(ring->dev->dev, addr)))
return 1;
switch (ring->type) {
case B43_DMA_30BIT:
if ((u64)addr + buffersize > (1ULL << 30))
goto address_error;
@@ -841,44 +846,44 @@ struct b43_dmaring *b43_setup_dmaring(st
b43_txhdr_size(dev),
GFP_KERNEL);
if (!ring->txhdr_cache)
goto err_kfree_meta;
/* test for ability to dma to txhdr_cache */
- dma_test = dma_map_single(dev->dev->dma_dev,
- ring->txhdr_cache,
- b43_txhdr_size(dev),
- DMA_TO_DEVICE);
+ dma_test = ssb_dma_map_single(dev->dev,
+ ring->txhdr_cache,
+ b43_txhdr_size(dev),
+ DMA_TO_DEVICE);
if (b43_dma_mapping_error(ring, dma_test,
b43_txhdr_size(dev), 1)) {
/* ugh realloc */
kfree(ring->txhdr_cache);
ring->txhdr_cache = kcalloc(ring->nr_slots,
b43_txhdr_size(dev),
GFP_KERNEL | GFP_DMA);
if (!ring->txhdr_cache)
goto err_kfree_meta;
- dma_test = dma_map_single(dev->dev->dma_dev,
- ring->txhdr_cache,
- b43_txhdr_size(dev),
- DMA_TO_DEVICE);
+ dma_test = ssb_dma_map_single(dev->dev,
+ ring->txhdr_cache,
+ b43_txhdr_size(dev),
+ DMA_TO_DEVICE);
if (b43_dma_mapping_error(ring, dma_test,
b43_txhdr_size(dev), 1)) {
b43err(dev->wl,
"TXHDR DMA allocation failed\n");
goto err_kfree_txhdr_cache;
}
}
- dma_unmap_single(dev->dev->dma_dev,
- dma_test, b43_txhdr_size(dev),
- DMA_TO_DEVICE);
+ ssb_dma_unmap_single(dev->dev,
+ dma_test, b43_txhdr_size(dev),
+ DMA_TO_DEVICE);
}
err = alloc_ringmemory(ring);
if (err)
goto err_kfree_txhdr_cache;
err = dmacontroller_setup(ring);
Index: wireless-testing/drivers/net/wireless/b43legacy/dma.c
===================================================================
--- wireless-testing.orig/drivers/net/wireless/b43legacy/dma.c 2008-06-20 02:06:45.000000000 +0200
+++ wireless-testing/drivers/net/wireless/b43legacy/dma.c 2008-06-20 02:08:02.000000000 +0200
@@ -390,59 +390,59 @@ dma_addr_t map_descbuffer(struct b43lega
size_t len,
int tx)
{
dma_addr_t dmaaddr;
if (tx)
- dmaaddr = dma_map_single(ring->dev->dev->dma_dev,
- buf, len,
- DMA_TO_DEVICE);
+ dmaaddr = ssb_dma_map_single(ring->dev->dev,
+ buf, len,
+ DMA_TO_DEVICE);
else
- dmaaddr = dma_map_single(ring->dev->dev->dma_dev,
- buf, len,
- DMA_FROM_DEVICE);
+ dmaaddr = ssb_dma_map_single(ring->dev->dev,
+ buf, len,
+ DMA_FROM_DEVICE);
return dmaaddr;
}
static inline
void unmap_descbuffer(struct b43legacy_dmaring *ring,
dma_addr_t addr,
size_t len,
int tx)
{
if (tx)
- dma_unmap_single(ring->dev->dev->dma_dev,
- addr, len,
- DMA_TO_DEVICE);
+ ssb_dma_unmap_single(ring->dev->dev,
+ addr, len,
+ DMA_TO_DEVICE);
else
- dma_unmap_single(ring->dev->dev->dma_dev,
- addr, len,
- DMA_FROM_DEVICE);
+ ssb_dma_unmap_single(ring->dev->dev,
+ addr, len,
+ DMA_FROM_DEVICE);
}
static inline
void sync_descbuffer_for_cpu(struct b43legacy_dmaring *ring,
dma_addr_t addr,
size_t len)
{
B43legacy_WARN_ON(ring->tx);
- dma_sync_single_for_cpu(ring->dev->dev->dma_dev,
- addr, len, DMA_FROM_DEVICE);
+ ssb_dma_sync_single_for_cpu(ring->dev->dev,
+ addr, len, DMA_FROM_DEVICE);
}
static inline
void sync_descbuffer_for_device(struct b43legacy_dmaring *ring,
dma_addr_t addr,
size_t len)
{
B43legacy_WARN_ON(ring->tx);
- dma_sync_single_for_device(ring->dev->dev->dma_dev,
- addr, len, DMA_FROM_DEVICE);
+ ssb_dma_sync_single_for_device(ring->dev->dev,
+ addr, len, DMA_FROM_DEVICE);
}
static inline
void free_descriptor_buffer(struct b43legacy_dmaring *ring,
struct b43legacy_dmadesc_meta *meta,
int irq_context)
@@ -455,32 +455,31 @@ void free_descriptor_buffer(struct b43le
meta->skb = NULL;
}
}
static int alloc_ringmemory(struct b43legacy_dmaring *ring)
{
- struct device *dma_dev = ring->dev->dev->dma_dev;
-
- ring->descbase = dma_alloc_coherent(dma_dev, B43legacy_DMA_RINGMEMSIZE,
- &(ring->dmabase), GFP_KERNEL);
+ /* GFP flags must match the flags in free_ringmemory()! */
+ ring->descbase = ssb_dma_alloc_consistent(ring->dev->dev,
+ B43legacy_DMA_RINGMEMSIZE,
+ &(ring->dmabase),
+ GFP_KERNEL);
if (!ring->descbase) {
b43legacyerr(ring->dev->wl, "DMA ringmemory allocation"
" failed\n");
return -ENOMEM;
}
memset(ring->descbase, 0, B43legacy_DMA_RINGMEMSIZE);
return 0;
}
static void free_ringmemory(struct b43legacy_dmaring *ring)
{
- struct device *dma_dev = ring->dev->dev->dma_dev;
-
- dma_free_coherent(dma_dev, B43legacy_DMA_RINGMEMSIZE,
- ring->descbase, ring->dmabase);
+ ssb_dma_free_consistent(ring->dev->dev, B43legacy_DMA_RINGMEMSIZE,
+ ring->descbase, ring->dmabase, GFP_KERNEL);
}
/* Reset the RX DMA channel */
static int b43legacy_dmacontroller_rx_reset(struct b43legacy_wldev *dev,
u16 mmio_base,
enum b43legacy_dmatype type)
@@ -586,13 +585,13 @@ static int b43legacy_dmacontroller_tx_re
/* Check if a DMA mapping address is invalid. */
static bool b43legacy_dma_mapping_error(struct b43legacy_dmaring *ring,
dma_addr_t addr,
size_t buffersize,
bool dma_to_device)
{
- if (unlikely(dma_mapping_error(addr)))
+ if (unlikely(ssb_dma_mapping_error(ring->dev->dev, addr)))
return 1;
switch (ring->type) {
case B43legacy_DMA_30BIT:
if ((u64)addr + buffersize > (1ULL << 30))
goto address_error;
@@ -891,39 +890,39 @@ struct b43legacy_dmaring *b43legacy_setu
sizeof(struct b43legacy_txhdr_fw3),
GFP_KERNEL);
if (!ring->txhdr_cache)
goto err_kfree_meta;
/* test for ability to dma to txhdr_cache */
- dma_test = dma_map_single(dev->dev->dma_dev, ring->txhdr_cache,
- sizeof(struct b43legacy_txhdr_fw3),
- DMA_TO_DEVICE);
+ dma_test = ssb_dma_map_single(dev->dev, ring->txhdr_cache,
+ sizeof(struct b43legacy_txhdr_fw3),
+ DMA_TO_DEVICE);
if (b43legacy_dma_mapping_error(ring, dma_test,
sizeof(struct b43legacy_txhdr_fw3), 1)) {
/* ugh realloc */
kfree(ring->txhdr_cache);
ring->txhdr_cache = kcalloc(nr_slots,
sizeof(struct b43legacy_txhdr_fw3),
GFP_KERNEL | GFP_DMA);
if (!ring->txhdr_cache)
goto err_kfree_meta;
- dma_test = dma_map_single(dev->dev->dma_dev,
+ dma_test = ssb_dma_map_single(dev->dev,
ring->txhdr_cache,
sizeof(struct b43legacy_txhdr_fw3),
DMA_TO_DEVICE);
if (b43legacy_dma_mapping_error(ring, dma_test,
sizeof(struct b43legacy_txhdr_fw3), 1))
goto err_kfree_txhdr_cache;
}
- dma_unmap_single(dev->dev->dma_dev,
- dma_test, sizeof(struct b43legacy_txhdr_fw3),
- DMA_TO_DEVICE);
+ ssb_dma_unmap_single(dev->dev, dma_test,
+ sizeof(struct b43legacy_txhdr_fw3),
+ DMA_TO_DEVICE);
}
ring->nr_slots = nr_slots;
ring->mmio_base = b43legacy_dmacontroller_base(type, controller_index);
ring->index = controller_index;
if (type == B43legacy_DMA_64BIT)
Index: wireless-testing/drivers/ssb/Kconfig
===================================================================
--- wireless-testing.orig/drivers/ssb/Kconfig 2008-06-20 02:04:56.000000000 +0200
+++ wireless-testing/drivers/ssb/Kconfig 2008-06-20 02:07:31.000000000 +0200
@@ -1,11 +1,11 @@
menu "Sonics Silicon Backplane"
config SSB_POSSIBLE
bool
- depends on HAS_IOMEM
+ depends on HAS_IOMEM && HAS_DMA
default y
config SSB
tristate "Sonics Silicon Backplane support"
depends on SSB_POSSIBLE
help
Index: wireless-testing/drivers/ssb/main.c
===================================================================
--- wireless-testing.orig/drivers/ssb/main.c 2008-06-20 02:04:56.000000000 +0200
+++ wireless-testing/drivers/ssb/main.c 2008-06-20 02:07:31.000000000 +0200
@@ -459,24 +459,21 @@ static int ssb_devices_register(struct s
switch (bus->bustype) {
case SSB_BUSTYPE_PCI:
#ifdef CONFIG_SSB_PCIHOST
sdev->irq = bus->host_pci->irq;
dev->parent = &bus->host_pci->dev;
- sdev->dma_dev = &bus->host_pci->dev;
#endif
break;
case SSB_BUSTYPE_PCMCIA:
#ifdef CONFIG_SSB_PCMCIAHOST
sdev->irq = bus->host_pcmcia->irq.AssignedIRQ;
dev->parent = &bus->host_pcmcia->dev;
- sdev->dma_dev = &bus->host_pcmcia->dev;
#endif
break;
case SSB_BUSTYPE_SSB:
- sdev->dma_dev = dev;
break;
}
sdev->dev = dev;
err = device_register(dev);
if (err) {
@@ -1153,42 +1150,88 @@ void ssb_device_disable(struct ssb_devic
EXPORT_SYMBOL(ssb_device_disable);
u32 ssb_dma_translation(struct ssb_device *dev)
{
switch (dev->bus->bustype) {
case SSB_BUSTYPE_SSB:
- case SSB_BUSTYPE_PCMCIA:
return 0;
case SSB_BUSTYPE_PCI:
return SSB_PCI_DMA;
+ default:
+ __ssb_dma_not_implemented(dev);
}
return 0;
}
EXPORT_SYMBOL(ssb_dma_translation);
-int ssb_dma_set_mask(struct ssb_device *ssb_dev, u64 mask)
+int ssb_dma_set_mask(struct ssb_device *dev, u64 mask)
{
- struct device *dma_dev = ssb_dev->dma_dev;
- int err = 0;
+ int err;
-#ifdef CONFIG_SSB_PCIHOST
- if (ssb_dev->bus->bustype == SSB_BUSTYPE_PCI) {
- err = pci_set_dma_mask(ssb_dev->bus->host_pci, mask);
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ err = pci_set_dma_mask(dev->bus->host_pci, mask);
if (err)
return err;
- err = pci_set_consistent_dma_mask(ssb_dev->bus->host_pci, mask);
+ err = pci_set_consistent_dma_mask(dev->bus->host_pci, mask);
return err;
+ case SSB_BUSTYPE_SSB:
+ return dma_set_mask(dev->dev, mask);
+ default:
+ __ssb_dma_not_implemented(dev);
}
-#endif
- dma_dev->coherent_dma_mask = mask;
- dma_dev->dma_mask = &dma_dev->coherent_dma_mask;
-
- return err;
+ return -ENOSYS;
}
EXPORT_SYMBOL(ssb_dma_set_mask);
+void * ssb_dma_alloc_consistent(struct ssb_device *dev, size_t size,
+ dma_addr_t *dma_handle, gfp_t gfp_flags)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ if (gfp_flags & GFP_DMA) {
+ /* Workaround: The PCI API does not support passing
+ * a GFP flag. */
+ return dma_alloc_coherent(&dev->bus->host_pci->dev,
+ size, dma_handle, gfp_flags);
+ }
+ return pci_alloc_consistent(dev->bus->host_pci, size, dma_handle);
+ case SSB_BUSTYPE_SSB:
+ return dma_alloc_coherent(dev->dev, size, dma_handle, gfp_flags);
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+ return NULL;
+}
+EXPORT_SYMBOL(ssb_dma_alloc_consistent);
+
+void ssb_dma_free_consistent(struct ssb_device *dev, size_t size,
+ void *vaddr, dma_addr_t dma_handle,
+ gfp_t gfp_flags)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ if (gfp_flags & GFP_DMA) {
+ /* Workaround: The PCI API does not support passing
+ * a GFP flag. */
+ dma_free_coherent(&dev->bus->host_pci->dev,
+ size, vaddr, dma_handle);
+ return;
+ }
+ pci_free_consistent(dev->bus->host_pci, size,
+ vaddr, dma_handle);
+ return;
+ case SSB_BUSTYPE_SSB:
+ dma_free_coherent(dev->dev, size, vaddr, dma_handle);
+ return;
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+}
+EXPORT_SYMBOL(ssb_dma_free_consistent);
+
int ssb_bus_may_powerdown(struct ssb_bus *bus)
{
struct ssb_chipcommon *cc;
int err = 0;
/* On buses where more than one core may be working
Index: wireless-testing/include/linux/ssb/ssb.h
===================================================================
--- wireless-testing.orig/include/linux/ssb/ssb.h 2008-06-20 02:04:56.000000000 +0200
+++ wireless-testing/include/linux/ssb/ssb.h 2008-06-20 02:07:31.000000000 +0200
@@ -134,15 +134,12 @@ struct __ssb_dev_wrapper {
struct ssb_device {
/* Having a copy of the ops pointer in each dev struct
* is an optimization. */
const struct ssb_bus_ops *ops;
struct device *dev;
- /* Pointer to the device that has to be used for
- * any DMA related operation. */
- struct device *dma_dev;
struct ssb_bus *bus;
struct ssb_device_id id;
u8 core_index;
unsigned int irq;
@@ -396,19 +393,157 @@ static inline void ssb_block_write(struc
{
dev->ops->block_write(dev, buffer, count, offset, reg_width);
}
#endif /* CONFIG_SSB_BLOCKIO */
+/* The SSB DMA API. Use this API for any DMA operation on the device.
+ * This API basically is a wrapper that calls the correct DMA API for
+ * the host device type the SSB device is attached to. */
+
/* Translation (routing) bits that need to be ORed to DMA
* addresses before they are given to a device. */
extern u32 ssb_dma_translation(struct ssb_device *dev);
#define SSB_DMA_TRANSLATION_MASK 0xC0000000
#define SSB_DMA_TRANSLATION_SHIFT 30
-extern int ssb_dma_set_mask(struct ssb_device *ssb_dev, u64 mask);
+extern int ssb_dma_set_mask(struct ssb_device *dev, u64 mask);
+
+extern void * ssb_dma_alloc_consistent(struct ssb_device *dev, size_t size,
+ dma_addr_t *dma_handle, gfp_t gfp_flags);
+extern void ssb_dma_free_consistent(struct ssb_device *dev, size_t size,
+ void *vaddr, dma_addr_t dma_handle,
+ gfp_t gfp_flags);
+
+static inline void __cold __ssb_dma_not_implemented(struct ssb_device *dev)
+{
+#ifdef CONFIG_SSB_DEBUG
+ printk(KERN_ERR "SSB: BUG! Calling DMA API for "
+ "unsupported bustype %d\n", dev->bus->bustype);
+#endif /* DEBUG */
+}
+
+static inline int ssb_dma_mapping_error(struct ssb_device *dev, dma_addr_t addr)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ return pci_dma_mapping_error(addr);
+ case SSB_BUSTYPE_SSB:
+ return dma_mapping_error(addr);
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+ return -ENOSYS;
+}
+
+static inline dma_addr_t ssb_dma_map_single(struct ssb_device *dev, void *p,
+ size_t size, enum dma_data_direction dir)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ return pci_map_single(dev->bus->host_pci, p, size, dir);
+ case SSB_BUSTYPE_SSB:
+ return dma_map_single(dev->dev, p, size, dir);
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+ return 0;
+}
+
+static inline void ssb_dma_unmap_single(struct ssb_device *dev, dma_addr_t dma_addr,
+ size_t size, enum dma_data_direction dir)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ pci_unmap_single(dev->bus->host_pci, dma_addr, size, dir);
+ return;
+ case SSB_BUSTYPE_SSB:
+ dma_unmap_single(dev->dev, dma_addr, size, dir);
+ return;
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+}
+
+static inline void ssb_dma_sync_single_for_cpu(struct ssb_device *dev,
+ dma_addr_t dma_addr,
+ size_t size,
+ enum dma_data_direction dir)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ pci_dma_sync_single_for_cpu(dev->bus->host_pci, dma_addr,
+ size, dir);
+ return;
+ case SSB_BUSTYPE_SSB:
+ dma_sync_single_for_cpu(dev->dev, dma_addr, size, dir);
+ return;
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+}
+
+static inline void ssb_dma_sync_single_for_device(struct ssb_device *dev,
+ dma_addr_t dma_addr,
+ size_t size,
+ enum dma_data_direction dir)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ pci_dma_sync_single_for_device(dev->bus->host_pci, dma_addr,
+ size, dir);
+ return;
+ case SSB_BUSTYPE_SSB:
+ dma_sync_single_for_device(dev->dev, dma_addr, size, dir);
+ return;
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+}
+
+static inline void ssb_dma_sync_single_range_for_cpu(struct ssb_device *dev,
+ dma_addr_t dma_addr,
+ unsigned long offset,
+ size_t size,
+ enum dma_data_direction dir)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ /* Just sync everything. That's all the PCI API can do. */
+ pci_dma_sync_single_for_cpu(dev->bus->host_pci, dma_addr,
+ offset + size, dir);
+ return;
+ case SSB_BUSTYPE_SSB:
+ dma_sync_single_range_for_cpu(dev->dev, dma_addr, offset,
+ size, dir);
+ return;
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+}
+
+static inline void ssb_dma_sync_single_range_for_device(struct ssb_device *dev,
+ dma_addr_t dma_addr,
+ unsigned long offset,
+ size_t size,
+ enum dma_data_direction dir)
+{
+ switch (dev->bus->bustype) {
+ case SSB_BUSTYPE_PCI:
+ /* Just sync everything. That's all the PCI API can do. */
+ pci_dma_sync_single_for_device(dev->bus->host_pci, dma_addr,
+ offset + size, dir);
+ return;
+ case SSB_BUSTYPE_SSB:
+ dma_sync_single_range_for_device(dev->dev, dma_addr, offset,
+ size, dir);
+ return;
+ default:
+ __ssb_dma_not_implemented(dev);
+ }
+}
#ifdef CONFIG_SSB_PCIHOST
/* PCI-host wrapper driver */
extern int ssb_pcihost_register(struct pci_driver *driver);
static inline void ssb_pcihost_unregister(struct pci_driver *driver)
On Saturday 21 June 2008 02:05:38 Larry Finger wrote:
> Is there a pre-requisite for this patch?
[PATCH] b43legacy: Fix possible NULL pointer dereference in DMA code
--
Greetings Michael.
Michael Buesch wrote:
> This is a rewrite of the DMA API for SSB devices.
> This is needed, because the old (non-existing) "API" made too many bad
> assumptions on the API of the host-bus (PCI).
> This introduces an almost complete SSB-DMA-API that maps to the lowlevel
> bus-API based on the bustype.
>
> Signed-off-by: Michael Buesch <[email protected]>
>
Is there a pre-requisite for this patch? I tried to apply it to
wireless-testing with version v2.6.26-rc6-10526-ga0d3022. Hunk 4 of
the patch for drivers/net/wireless/b43legacy/dma.c failed due to a
line missing from the original source:
> @@ -891,39 +890,39 @@ struct b43legacy_dmaring *b43legacy_setu
--snip--
> + ssb_dma_unmap_single(dev->dev, dma_test,
> + sizeof(struct b43legacy_txhdr_fw3),
> + DMA_TO_DEVICE);
> }
>
ring->dev = dev; <-------------- This line missing.
> ring->nr_slots = nr_slots;
> ring->mmio_base = b43legacy_dmacontroller_base(type, controller_index);
> ring->index = controller_index;
> if (type == B43legacy_DMA_64BIT)
Larry
On Friday 20 June 2008 11:50:29 Michael Buesch wrote:
> This is a rewrite of the DMA API for SSB devices.
> This is needed, because the old (non-existing) "API" made too many bad
> assumptions on the API of the host-bus (PCI).
> This introduces an almost complete SSB-DMA-API that maps to the lowlevel
> bus-API based on the bustype.
>
> Signed-off-by: Michael Buesch <[email protected]>
>
> ---
>
> John, please merge for 2.6.27
> This is one huge patch, because it's not really possible to split this
> without breaking bisect.
Note that this patch will slightly clash with another DMA update
that's going into linux-next.
This is trivial to resolve, however. wiggle will probably be able to
resolve most, if not all, of them.
> static bool b43_dma_mapping_error(struct b43_dmaring *ring,
> dma_addr_t addr,
> size_t buffersize, bool dma_to_device)
> {
> - if (unlikely(dma_mapping_error(addr)))
> + if (unlikely(ssb_dma_mapping_error(ring->dev->dev, addr)))
> return 1;
These calls will clash. The patch in linux-next introduced an additional
dev parameter to dma_mapping_error.
The fix is obvious. Remove the clashing line and add the ssb_dma_mapping_error
call exactly as shown above.
> +static inline int ssb_dma_mapping_error(struct ssb_device *dev, dma_addr_t addr)
> +{
> + switch (dev->bus->bustype) {
> + case SSB_BUSTYPE_PCI:
> + return pci_dma_mapping_error(addr);
> + case SSB_BUSTYPE_SSB:
> + return dma_mapping_error(addr);
> + default:
> + __ssb_dma_not_implemented(dev);
> + }
> + return -ENOSYS;
> +}
The patch in linux-next will require a dev parameter to the lowlevel
functions. Simply do it this way:
pci_dma_mapping_error(dev->bus->host_pci, addr);
and
dma_mapping_error(dev->dev, addr);
--
Greetings Michael.
On Tuesday 08 July 2008 17:58:58 Steve Brown wrote:
> I'll post the problem to linux-mips.
Please do so. This really is a DMA API issue. It's either using the masks
incorrectly, or ot should provide a function to set the coherent mask.
SSB does use The Right (TM) API. So it should not break, in my opinion.
I'm fully aware, that there is no non-PCI function that sets the coherent
mask. But I really expected that nobody is using it for non-PCI devices
either.
--
Greetings Michael.
On Thursday 03 July 2008 18:30:18 Felipe Maya wrote:
> Hi, I have applied this patch and i tested on a wl500gpv2 (BCM5354). The
> following part disable the ethernet device. If this part of the patch is
> restored the device work ok.
Well, I would say your architecture is pretty broken then.
This patch replaces an incorrect manual DMA mask setting by
a call to the official API.
Please report this to your archtitecture maintainer.
> > -int ssb_dma_set_mask(struct ssb_device *ssb_dev, u64 mask)
> > +int ssb_dma_set_mask(struct ssb_device *dev, u64 mask)
> > {
> > - struct device *dma_dev = ssb_dev->dma_dev;
> > - int err = 0;
> > + int err;
> >
> > -#ifdef CONFIG_SSB_PCIHOST
> > - if (ssb_dev->bus->bustype == SSB_BUSTYPE_PCI) {
> > - err = pci_set_dma_mask(ssb_dev->bus->host_pci, mask);
> > + switch (dev->bus->bustype) {
> > + case SSB_BUSTYPE_PCI:
> > + err = pci_set_dma_mask(dev->bus->host_pci, mask);
> > if (err)
> > return err;
> > - err = pci_set_consistent_dma_mask(ssb_dev->bus->host_pci, mask);
> > + err = pci_set_consistent_dma_mask(dev->bus->host_pci, mask);
> > return err;
> > + case SSB_BUSTYPE_SSB:
> > + return dma_set_mask(dev->dev, mask);
> > + default:
> > + __ssb_dma_not_implemented(dev);
> > }
> > -#endif
> > - dma_dev->coherent_dma_mask = mask;
> > - dma_dev->dma_mask = &dma_dev->coherent_dma_mask;
> > -
> > - return err;
> > + return -ENOSYS;
> > }
> > EXPORT_SYMBOL(ssb_dma_set_mask);
>
>
>
>
--
Greetings Michael.
Hi, I have applied this patch and i tested on a wl500gpv2 (BCM5354). The
following part disable the ethernet device. If this part of the patch is
restored the device work ok.
>
> -int ssb_dma_set_mask(struct ssb_device *ssb_dev, u64 mask)
> +int ssb_dma_set_mask(struct ssb_device *dev, u64 mask)
> {
> - struct device *dma_dev = ssb_dev->dma_dev;
> - int err = 0;
> + int err;
>
> -#ifdef CONFIG_SSB_PCIHOST
> - if (ssb_dev->bus->bustype == SSB_BUSTYPE_PCI) {
> - err = pci_set_dma_mask(ssb_dev->bus->host_pci, mask);
> + switch (dev->bus->bustype) {
> + case SSB_BUSTYPE_PCI:
> + err = pci_set_dma_mask(dev->bus->host_pci, mask);
> if (err)
> return err;
> - err = pci_set_consistent_dma_mask(ssb_dev->bus->host_pci, mask);
> + err = pci_set_consistent_dma_mask(dev->bus->host_pci, mask);
> return err;
> + case SSB_BUSTYPE_SSB:
> + return dma_set_mask(dev->dev, mask);
> + default:
> + __ssb_dma_not_implemented(dev);
> }
> -#endif
> - dma_dev->coherent_dma_mask = mask;
> - dma_dev->dma_mask = &dma_dev->coherent_dma_mask;
> -
> - return err;
> + return -ENOSYS;
> }
> EXPORT_SYMBOL(ssb_dma_set_mask);
On Thursday 03 July 2008 22:52:35 Michael Buesch wrote:
> On Thursday 03 July 2008 18:30:18 Felipe Maya wrote:
> > Hi, I have applied this patch and i tested on a wl500gpv2 (BCM5354). The
> > following part disable the ethernet device. If this part of the patch is
> > restored the device work ok.
>
>
> Well, I would say your architecture is pretty broken then.
> This patch replaces an incorrect manual DMA mask setting by
> a call to the official API.
> Please report this to your archtitecture maintainer.
Ah and btw, the bug most likely is that dma_set_mask does not
set the dev->coherent_dma_mask or something like that
while the dma remapping function require it to be set (I could be wrong,
this is just guesswork). In any case, I think the bug really is in the
arch DMA implementation and we must not workaround that in the ssb code.
So I will reject a patch that adds the
dev->coherent_dma_mask = mask;
dma_dev->dma_mask = &dma_dev->coherent_dma_mask;
stuff back. dma_set_mask should set all required masks for the dma_...
remapping functions. If it doesn't, it's either a bug in dma_set_mask
or in the remapping functions using the wrong mask.
We need to fix it correctly this time, as the whole purpose of
my SSB patch was the get the DMA remapping _right_ so it doesn't break
every two months if somebody changes some arch internal assumption.
So please do not to send me any SSB patch that works around the bug. ;)
>
> > > -int ssb_dma_set_mask(struct ssb_device *ssb_dev, u64 mask)
> > > +int ssb_dma_set_mask(struct ssb_device *dev, u64 mask)
> > > {
> > > - struct device *dma_dev = ssb_dev->dma_dev;
> > > - int err = 0;
> > > + int err;
> > >
> > > -#ifdef CONFIG_SSB_PCIHOST
> > > - if (ssb_dev->bus->bustype == SSB_BUSTYPE_PCI) {
> > > - err = pci_set_dma_mask(ssb_dev->bus->host_pci, mask);
> > > + switch (dev->bus->bustype) {
> > > + case SSB_BUSTYPE_PCI:
> > > + err = pci_set_dma_mask(dev->bus->host_pci, mask);
> > > if (err)
> > > return err;
> > > - err = pci_set_consistent_dma_mask(ssb_dev->bus->host_pci, mask);
> > > + err = pci_set_consistent_dma_mask(dev->bus->host_pci, mask);
> > > return err;
> > > + case SSB_BUSTYPE_SSB:
> > > + return dma_set_mask(dev->dev, mask);
> > > + default:
> > > + __ssb_dma_not_implemented(dev);
> > > }
> > > -#endif
> > > - dma_dev->coherent_dma_mask = mask;
> > > - dma_dev->dma_mask = &dma_dev->coherent_dma_mask;
> > > -
> > > - return err;
> > > + return -ENOSYS;
> > > }
> > > EXPORT_SYMBOL(ssb_dma_set_mask);
> >
> >
> >
> >
>
>
>
--
Greetings Michael.
Michael Buesch wrote:
> On Thursday 03 July 2008 18:30:18 Felipe Maya wrote:
>
>> Hi, I have applied this patch and i tested on a wl500gpv2 (BCM5354). The
>> following part disable the ethernet device. If this part of the patch is
>> restored the device work ok.
>>
>
>
> Well, I would say your architecture is pretty broken then.
> This patch replaces an incorrect manual DMA mask setting by
> a call to the official API.
> Please report this to your archtitecture maintainer.
>
>
>
This also breaks at least b44 on my wl500gpv1.
with wireless-testing rc9-wl tree:
b44.c:v2.0
b44 ssb0:0: Required 30BIT DMA mask unsupported by the system.
b44: probe of ssb0:0 failed with error -5
b44 ssb0:1: Required 30BIT DMA mask unsupported by the system.
b44: probe of ssb0:1 failed with error -5
and reverting the patch:
b44.c:v2.0
eth0: Broadcom 44xx/47xx 10/100BaseT Ethernet 00:17:31:ba:ec:35
eth1: Broadcom 44xx/47xx 10/100BaseT Ethernet 40:10:18:00:00:2d
The problem appears to be a missing non-pci equivalent of
pci_set_consistent_dma_mask. Maybe that's why the mask was set manually.
I can not find anything that sets coherent_dma_mask for other than a pci
device. If somebody can confirm my (mis)understanding of this, I'll post
the problem to linux-mips.
Steve