2019-08-19 16:34:22

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 00/17] Use MFD framework for SGI IOC3 drivers

GI IOC3 ASIC includes support for ethernet, PS2 keyboard/mouse,
NIC (number in a can), GPIO and a byte bus. By attaching a
SuperIO chip to it, it also supports serial lines and a parallel
port. The chip is used on a variety of SGI systems with different
configurations. This patchset moves code out of the network driver,
which doesn't belong there, into its new place a MFD driver and
specific platform drivers for the different subfunctions.

Changes in v5:
- requested by Jakub I've splitted ioc3 ethernet driver changes into
more steps to make the transition more visible; on the way there
I've "checkpatched" the driver and reduced code reorderings
- dropped all uint16_t and uint32_t
- added nvmem API extension to the documenation file
- changed to use request_irq/free_irq in serio driver
- removed wrong kfree() in serio error path

Changes in v4:
- added w1 drivers to the series after merge in 5.3 failed because
of no response from maintainer and other parts of this series
won't work without that drivers
- moved ip30 systemboard support to the ip30 series, which will
deal with rtc oddity Lee found
- converted to use devm_platform_ioremap_resource
- use PLATFORM_DEVID_AUTO for serial, ethernet and serio in mfd driver
- fixed reverse christmas order in ioc3-eth.c
- formating issue found by Lee
- re-worked irq request/free in serio driver to avoid crashes during
probe/remove

Changes in v3:
- use 1-wire subsystem for handling proms
- pci-xtalk driver uses prom information to create PCI subsystem
ids for use in MFD driver
- changed MFD driver to only use static declared mfd_cells
- added IP30 system board setup to MFD driver
- mac address is now read from ioc3-eth driver with nvmem framework

Changes in v2:
- fixed issue in ioc3kbd.c reported by Dmitry Torokhov
- merged IP27 RTC removal and 8250 serial driver addition into
main MFD patch to keep patches bisectable

Thomas Bogendoerfer (17):
w1: add 1-wire master driver for IP block found in SGI ASICs
w1: add DS2501, DS2502, DS2505 EPROM device driver
nvmem: core: add nvmem_device_find
MIPS: PCI: refactor ioc3 special handling
MIPS: PCI: use information from 1-wire PROM for IOC3 detection
MIPS: SGI-IP27: remove ioc3 ethernet init
MIPS: SGI-IP27: restructure ioc3 register access
net: sgi: ioc3-eth: remove checkpatch errors/warning
net: sgi: ioc3-eth: use defines for constants dealing with desc rings
net: sgi: ioc3-eth: rework skb rx handling
net: sgi: ioc3-eth: no need to stop queue set_multicast_list
net: sgi: ioc3-eth: use dma-direct for dma allocations
net: sgi: ioc3-eth: use csum_fold
net: sgi: ioc3-eth: Fix IPG settings
mfd: ioc3: Add driver for SGI IOC3 chip
MIPS: SGI-IP27: fix readb/writeb addressing
Input: add IOC3 serio driver

Documentation/driver-api/nvmem.rst | 2 +
arch/mips/include/asm/mach-ip27/mangle-port.h | 4 +-
arch/mips/include/asm/pci/bridge.h | 1 +
arch/mips/include/asm/sn/ioc3.h | 364 +++----
arch/mips/pci/pci-xtalk-bridge.c | 296 ++++--
arch/mips/sgi-ip27/ip27-console.c | 5 +-
arch/mips/sgi-ip27/ip27-init.c | 13 -
arch/mips/sgi-ip27/ip27-timer.c | 20 -
arch/mips/sgi-ip27/ip27-xtalk.c | 38 +-
drivers/input/serio/Kconfig | 10 +
drivers/input/serio/Makefile | 1 +
drivers/input/serio/ioc3kbd.c | 160 +++
drivers/mfd/Kconfig | 13 +
drivers/mfd/Makefile | 1 +
drivers/mfd/ioc3.c | 586 +++++++++++
drivers/net/ethernet/sgi/Kconfig | 4 +-
drivers/net/ethernet/sgi/ioc3-eth.c | 1405 +++++++++----------------
drivers/nvmem/core.c | 62 +-
drivers/rtc/rtc-m48t35.c | 11 +
drivers/tty/serial/8250/8250_ioc3.c | 98 ++
drivers/tty/serial/8250/Kconfig | 11 +
drivers/tty/serial/8250/Makefile | 1 +
drivers/w1/masters/Kconfig | 9 +
drivers/w1/masters/Makefile | 1 +
drivers/w1/masters/sgi_w1.c | 130 +++
drivers/w1/slaves/Kconfig | 6 +
drivers/w1/slaves/Makefile | 1 +
drivers/w1/slaves/w1_ds250x.c | 293 ++++++
include/linux/nvmem-consumer.h | 9 +
include/linux/platform_data/sgi-w1.h | 15 +
include/linux/w1.h | 2 +
31 files changed, 2266 insertions(+), 1306 deletions(-)
create mode 100644 drivers/input/serio/ioc3kbd.c
create mode 100644 drivers/mfd/ioc3.c
create mode 100644 drivers/tty/serial/8250/8250_ioc3.c
create mode 100644 drivers/w1/masters/sgi_w1.c
create mode 100644 drivers/w1/slaves/w1_ds250x.c
create mode 100644 include/linux/platform_data/sgi-w1.h

--
2.13.7


2019-08-19 16:34:27

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 13/17] net: sgi: ioc3-eth: use csum_fold

replace open coded checksum folding by csum_fold.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 6 +-----
1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 647e3926bd71..56392f0b01de 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -1378,16 +1378,12 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
/* Sum up dest addr, src addr and protocol */
ehsum = eh[0] + eh[1] + eh[2] + eh[3] + eh[4] + eh[5] + eh[6];

- /* Fold ehsum. can't use csum_fold which negates also ... */
- ehsum = (ehsum & 0xffff) + (ehsum >> 16);
- ehsum = (ehsum & 0xffff) + (ehsum >> 16);
-
/* Skip IP header; it's sum is always zero and was
* already filled in by ip_output.c
*/
csum = csum_tcpudp_nofold(ih->saddr, ih->daddr,
ih->tot_len - (ih->ihl << 2),
- proto, 0xffff ^ ehsum);
+ proto, csum_fold(ehsum));

csum = (csum & 0xffff) + (csum >> 16); /* Fold again */
csum = (csum & 0xffff) + (csum >> 16);
--
2.13.7

2019-08-19 16:34:44

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 12/17] net: sgi: ioc3-eth: use dma-direct for dma allocations

Replace the homegrown DMA memory allocation, which only works on
SGI-IP27 machines, with the generic dma allocations.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 107 ++++++++++++++++++++++++++----------
1 file changed, 77 insertions(+), 30 deletions(-)

diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 7f85a3bfef14..647e3926bd71 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -38,7 +38,6 @@
#include <linux/ip.h>
#include <linux/tcp.h>
#include <linux/udp.h>
-#include <linux/dma-mapping.h>
#include <linux/gfp.h>

#ifdef CONFIG_SERIAL_8250
@@ -51,6 +50,8 @@
#include <linux/etherdevice.h>
#include <linux/ethtool.h>
#include <linux/skbuff.h>
+#include <linux/dma-direct.h>
+
#include <net/ip.h>

#include <asm/byteorder.h>
@@ -66,10 +67,12 @@
#define RX_BUFFS 64
#define RX_RING_ENTRIES 512 /* fixed in hardware */
#define RX_RING_MASK (RX_RING_ENTRIES - 1)
+#define RX_RING_SIZE (RX_RING_ENTRIES * sizeof(u64))

/* 128 TX buffers (not tunable) */
#define TX_RING_ENTRIES 128
#define TX_RING_MASK (TX_RING_ENTRIES - 1)
+#define TX_RING_SIZE (TX_RING_ENTRIES * sizeof(struct ioc3_etxd))

/* BEWARE: The IOC3 documentation documents the size of rx buffers as
* 1644 while it's actually 1664. This one was nasty to track down...
@@ -84,9 +87,12 @@
struct ioc3_private {
struct ioc3_ethregs *regs;
struct ioc3 *all_regs;
+ struct device *dma_dev;
u32 *ssram;
unsigned long *rxr; /* pointer to receiver ring */
struct ioc3_etxd *txr;
+ dma_addr_t rxr_dma;
+ dma_addr_t txr_dma;
struct sk_buff *rx_skbs[RX_RING_ENTRIES];
struct sk_buff *tx_skbs[TX_RING_ENTRIES];
int rx_ci; /* RX consumer index */
@@ -116,18 +122,22 @@ static void ioc3_init(struct net_device *dev);
static const char ioc3_str[] = "IOC3 Ethernet";
static const struct ethtool_ops ioc3_ethtool_ops;

-static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
+#ifdef CONFIG_PCI_XTALK_BRIDGE
+static inline unsigned long ioc3_map(dma_addr_t addr, unsigned long attr)
{
-#ifdef CONFIG_SGI_IP27
- vdev <<= 57; /* Shift to PCI64_ATTR_VIRTUAL */
+ return (addr & ~PCI64_ATTR_BAR) | attr;
+}

- return vdev | (0xaUL << PCI64_ATTR_TARG_SHFT) | PCI64_ATTR_PREF |
- ((unsigned long)ptr & TO_PHYS_MASK);
+#define ERBAR_VAL (ERBAR_BARRIER_BIT << ERBAR_RXBARR_SHIFT)
#else
- return virt_to_bus(ptr);
-#endif
+static inline unsigned long ioc3_map(dma_addr_t addr, unsigned long attr)
+{
+ return addr;
}

+#define ERBAR_VAL 0
+#endif
+
#define IOC3_SIZE 0x100000

static inline u32 mcr_pack(u32 pulse, u32 sample)
@@ -494,6 +504,7 @@ static inline void ioc3_rx(struct net_device *dev)
int rx_entry, n_entry, len;
struct ioc3_erxbuf *rxb;
unsigned long *rxr;
+ dma_addr_t d;
u32 w0, err;

rxr = ip->rxr; /* Ring base */
@@ -550,7 +561,9 @@ static inline void ioc3_rx(struct net_device *dev)
dev->stats.rx_frame_errors++;
next:
ip->rx_skbs[n_entry] = new_skb;
- rxr[n_entry] = cpu_to_be64(ioc3_map(rxb, 1));
+ d = dma_map_single(ip->dma_dev, rxb, RX_BUF_SIZE,
+ DMA_FROM_DEVICE);
+ rxr[n_entry] = cpu_to_be64(ioc3_map(d, PCI64_ATTR_BAR));
rxb->w0 = 0; /* Clear valid flag */
n_entry = (n_entry + 1) & RX_RING_MASK; /* Update erpir */

@@ -754,6 +767,26 @@ static inline void ioc3_clean_rx_ring(struct ioc3_private *ip)
}
}

+static inline void ioc3_tx_unmap(struct ioc3_private *ip, int entry)
+{
+ struct ioc3_etxd *desc;
+ u32 cmd, bufcnt, len;
+
+ desc = &ip->txr[entry];
+ cmd = be32_to_cpu(desc->cmd);
+ bufcnt = be32_to_cpu(desc->bufcnt);
+ if (cmd & ETXD_B1V) {
+ len = (bufcnt & ETXD_B1CNT_MASK) >> ETXD_B1CNT_SHIFT;
+ dma_unmap_single(ip->dma_dev, be64_to_cpu(desc->p1),
+ len, DMA_TO_DEVICE);
+ }
+ if (cmd & ETXD_B2V) {
+ len = (bufcnt & ETXD_B2CNT_MASK) >> ETXD_B2CNT_SHIFT;
+ dma_unmap_single(ip->dma_dev, be64_to_cpu(desc->p2),
+ len, DMA_TO_DEVICE);
+ }
+}
+
static inline void ioc3_clean_tx_ring(struct ioc3_private *ip)
{
struct sk_buff *skb;
@@ -762,6 +795,7 @@ static inline void ioc3_clean_tx_ring(struct ioc3_private *ip)
for (i = 0; i < TX_RING_ENTRIES; i++) {
skb = ip->tx_skbs[i];
if (skb) {
+ ioc3_tx_unmap(ip, i);
ip->tx_skbs[i] = NULL;
dev_kfree_skb_any(skb);
}
@@ -778,7 +812,8 @@ static void ioc3_free_rings(struct ioc3_private *ip)

if (ip->txr) {
ioc3_clean_tx_ring(ip);
- free_pages((unsigned long)ip->txr, 2);
+ dma_direct_free_pages(ip->dma_dev, TX_RING_SIZE, ip->txr,
+ ip->txr_dma, 0);
ip->txr = NULL;
}

@@ -788,12 +823,17 @@ static void ioc3_free_rings(struct ioc3_private *ip)

while (n_entry != rx_entry) {
skb = ip->rx_skbs[n_entry];
- if (skb)
+ if (skb) {
+ dma_unmap_single(ip->dma_dev,
+ be64_to_cpu(ip->rxr[n_entry]),
+ RX_BUF_SIZE, DMA_FROM_DEVICE);
dev_kfree_skb_any(skb);
+ }

n_entry = (n_entry + 1) & RX_RING_MASK;
}
- free_page((unsigned long)ip->rxr);
+ dma_direct_free_pages(ip->dma_dev, RX_RING_SIZE, ip->rxr,
+ ip->rxr_dma, 0);
ip->rxr = NULL;
}
}
@@ -801,16 +841,19 @@ static void ioc3_free_rings(struct ioc3_private *ip)
static void ioc3_alloc_rings(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3_erxbuf *rxb;
unsigned long *rxr;
+ dma_addr_t rxb;
int i;

if (!ip->rxr) {
/* Allocate and initialize rx ring. 4kb = 512 entries */
- ip->rxr = (unsigned long *)get_zeroed_page(GFP_ATOMIC);
+ ip->rxr = dma_direct_alloc_pages(ip->dma_dev, RX_RING_SIZE,
+ &ip->rxr_dma, GFP_ATOMIC, 0);
rxr = ip->rxr;
- if (!rxr)
+ if (!rxr) {
pr_err("%s: get_zeroed_page() failed!\n", __func__);
+ return;
+ }

/* Now the rx buffers. The RX ring may be larger but
* we only allocate 16 buffers for now. Need to tune
@@ -828,8 +871,9 @@ static void ioc3_alloc_rings(struct net_device *dev)

ip->rx_skbs[i] = skb;

- rxb = (struct ioc3_erxbuf *)skb->data;
- rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
+ rxb = dma_map_single(ip->dma_dev, skb->data,
+ RX_BUF_SIZE, DMA_BIDIRECTIONAL);
+ rxr[i] = cpu_to_be64(ioc3_map(rxb, PCI64_ATTR_BAR));
skb_reserve(skb, RX_OFFSET);
}
ip->rx_ci = 0;
@@ -838,7 +882,9 @@ static void ioc3_alloc_rings(struct net_device *dev)

if (!ip->txr) {
/* Allocate and initialize tx rings. 16kb = 128 bufs. */
- ip->txr = (struct ioc3_etxd *)__get_free_pages(GFP_KERNEL, 2);
+ ip->txr = dma_direct_alloc_pages(ip->dma_dev, TX_RING_SIZE,
+ &ip->txr_dma,
+ GFP_KERNEL | __GFP_ZERO, 0);
if (!ip->txr)
pr_err("%s: __get_free_pages() failed!\n", __func__);
ip->tx_pi = 0;
@@ -859,13 +905,13 @@ static void ioc3_init_rings(struct net_device *dev)
ioc3_clean_tx_ring(ip);

/* Now the rx ring base, consume & produce registers. */
- ring = ioc3_map(ip->rxr, 0);
+ ring = ioc3_map(ip->rxr_dma, PCI64_ATTR_PREC);
writel(ring >> 32, &regs->erbr_h);
writel(ring & 0xffffffff, &regs->erbr_l);
writel(ip->rx_ci << 3, &regs->ercir);
writel((ip->rx_pi << 3) | ERPIR_ARM, &regs->erpir);

- ring = ioc3_map(ip->txr, 0);
+ ring = ioc3_map(ip->txr_dma, PCI64_ATTR_PREC);

ip->txqlen = 0; /* nothing queued */

@@ -915,13 +961,7 @@ static void ioc3_init(struct net_device *dev)
readl(&regs->emcr);

/* Misc registers */
-#ifdef CONFIG_SGI_IP27
- /* Barrier on last store */
- writel(PCI64_ATTR_BAR >> 32, &regs->erbar);
-#else
- /* Let PCI API get it right */
- writel(0, &regs->erbar);
-#endif
+ writel(ERBAR_VAL, &regs->erbar);
readl(&regs->etcdc); /* Clear on read */
writel(15, &regs->ercsr); /* RX low watermark */
writel(0, &regs->ertr); /* Interrupt immediately */
@@ -1187,6 +1227,7 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)

ip = netdev_priv(dev);
ip->dev = dev;
+ ip->dma_dev = &pdev->dev;

dev->irq = pdev->irq;

@@ -1386,18 +1427,24 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
unsigned long b2 = (data | 0x3fffUL) + 1UL;
unsigned long s1 = b2 - data;
unsigned long s2 = data + len - b2;
+ dma_addr_t d;

desc->cmd = cpu_to_be32(len | ETXD_INTWHENDONE |
ETXD_B1V | ETXD_B2V | w0);
desc->bufcnt = cpu_to_be32((s1 << ETXD_B1CNT_SHIFT) |
(s2 << ETXD_B2CNT_SHIFT));
- desc->p1 = cpu_to_be64(ioc3_map(skb->data, 1));
- desc->p2 = cpu_to_be64(ioc3_map((void *)b2, 1));
+ d = dma_map_single(ip->dma_dev, skb->data, s1, DMA_TO_DEVICE);
+ desc->p1 = cpu_to_be64(ioc3_map(d, PCI64_ATTR_PREF));
+ d = dma_map_single(ip->dma_dev, (void *)b2, s1, DMA_TO_DEVICE);
+ desc->p2 = cpu_to_be64(ioc3_map(d, PCI64_ATTR_PREF));
} else {
+ dma_addr_t d;
+
/* Normal sized packet that doesn't cross a page boundary. */
desc->cmd = cpu_to_be32(len | ETXD_INTWHENDONE | ETXD_B1V | w0);
desc->bufcnt = cpu_to_be32(len << ETXD_B1CNT_SHIFT);
- desc->p1 = cpu_to_be64(ioc3_map(skb->data, 1));
+ d = dma_map_single(ip->dma_dev, skb->data, len, DMA_TO_DEVICE);
+ desc->p1 = cpu_to_be64(ioc3_map(d, PCI64_ATTR_PREF));
}

mb(); /* make sure all descriptor changes are visible */
--
2.13.7

2019-08-19 16:34:53

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 10/17] net: sgi: ioc3-eth: rework skb rx handling

Buffers alloacted by alloc_skb() are already cache aligned so there
is no need for an extra align done by ioc3_alloc_skb. And instead
of skb_put/skb_trim simply use one skb_put after frame size is known
during receive.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 50 ++++++++-----------------------------
1 file changed, 11 insertions(+), 39 deletions(-)

diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index c875640926d6..d862f28887f9 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -11,7 +11,6 @@
*
* To do:
*
- * o Handle allocation failures in ioc3_alloc_skb() more gracefully.
* o Handle allocation failures in ioc3_init_rings().
* o Use prefetching for large packets. What is a good lower limit for
* prefetching?
@@ -72,6 +71,12 @@
#define TX_RING_ENTRIES 128
#define TX_RING_MASK (TX_RING_ENTRIES - 1)

+/* BEWARE: The IOC3 documentation documents the size of rx buffers as
+ * 1644 while it's actually 1664. This one was nasty to track down...
+ */
+#define RX_OFFSET 10
+#define RX_BUF_SIZE 1664
+
#define ETCSR_FD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
#define ETCSR_HD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)

@@ -111,31 +116,6 @@ static void ioc3_init(struct net_device *dev);
static const char ioc3_str[] = "IOC3 Ethernet";
static const struct ethtool_ops ioc3_ethtool_ops;

-/* We use this to acquire receive skb's that we can DMA directly into. */
-
-#define IOC3_CACHELINE 128UL
-
-static inline unsigned long aligned_rx_skb_addr(unsigned long addr)
-{
- return (~addr + 1) & (IOC3_CACHELINE - 1UL);
-}
-
-static inline struct sk_buff *ioc3_alloc_skb(unsigned long length,
- unsigned int gfp_mask)
-{
- struct sk_buff *skb;
-
- skb = alloc_skb(length + IOC3_CACHELINE - 1, gfp_mask);
- if (likely(skb)) {
- int offset = aligned_rx_skb_addr((unsigned long)skb->data);
-
- if (offset)
- skb_reserve(skb, offset);
- }
-
- return skb;
-}
-
static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
{
#ifdef CONFIG_SGI_IP27
@@ -148,12 +128,6 @@ static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
#endif
}

-/* BEWARE: The IOC3 documentation documents the size of rx buffers as
- * 1644 while it's actually 1664. This one was nasty to track down ...
- */
-#define RX_OFFSET 10
-#define RX_BUF_ALLOC_SIZE (1664 + RX_OFFSET + IOC3_CACHELINE)
-
#define IOC3_SIZE 0x100000

static inline u32 mcr_pack(u32 pulse, u32 sample)
@@ -534,10 +508,10 @@ static inline void ioc3_rx(struct net_device *dev)
err = be32_to_cpu(rxb->err); /* It's valid ... */
if (err & ERXBUF_GOODPKT) {
len = ((w0 >> ERXBUF_BYTECNT_SHIFT) & 0x7ff) - 4;
- skb_trim(skb, len);
+ skb_put(skb, len);
skb->protocol = eth_type_trans(skb, dev);

- new_skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
+ new_skb = alloc_skb(RX_BUF_SIZE, GFP_ATOMIC);
if (!new_skb) {
/* Ouch, drop packet and just recycle packet
* to keep the ring filled.
@@ -546,6 +520,7 @@ static inline void ioc3_rx(struct net_device *dev)
new_skb = skb;
goto next;
}
+ new_skb->dev = dev;

if (likely(dev->features & NETIF_F_RXCSUM))
ioc3_tcpudp_checksum(skb,
@@ -556,8 +531,6 @@ static inline void ioc3_rx(struct net_device *dev)

ip->rx_skbs[rx_entry] = NULL; /* Poison */

- /* Because we reserve afterwards. */
- skb_put(new_skb, (1664 + RX_OFFSET));
rxb = (struct ioc3_erxbuf *)new_skb->data;
skb_reserve(new_skb, RX_OFFSET);

@@ -846,16 +819,15 @@ static void ioc3_alloc_rings(struct net_device *dev)
for (i = 0; i < RX_BUFFS; i++) {
struct sk_buff *skb;

- skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
+ skb = alloc_skb(RX_BUF_SIZE, GFP_ATOMIC);
if (!skb) {
show_free_areas(0, NULL);
continue;
}
+ skb->dev = dev;

ip->rx_skbs[i] = skb;

- /* Because we reserve afterwards. */
- skb_put(skb, (1664 + RX_OFFSET));
rxb = (struct ioc3_erxbuf *)skb->data;
rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
skb_reserve(skb, RX_OFFSET);
--
2.13.7

2019-08-19 16:34:59

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 09/17] net: sgi: ioc3-eth: use defines for constants dealing with desc rings

Descriptor ring sizes of the IOC3 are more or less fixed size. To
make clearer where there is a relation to ring sizes use defines.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 42 +++++++++++++++++++++----------------
1 file changed, 24 insertions(+), 18 deletions(-)

diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index e81e5bb37ac6..c875640926d6 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -61,10 +61,16 @@
#include <asm/sn/ioc3.h>
#include <asm/pci/bridge.h>

-/* 64 RX buffers. This is tunable in the range of 16 <= x < 512. The
- * value must be a power of two.
+/* Number of RX buffers. This is tunable in the range of 16 <= x < 512.
+ * The value must be a power of two.
*/
-#define RX_BUFFS 64
+#define RX_BUFFS 64
+#define RX_RING_ENTRIES 512 /* fixed in hardware */
+#define RX_RING_MASK (RX_RING_ENTRIES - 1)
+
+/* 128 TX buffers (not tunable) */
+#define TX_RING_ENTRIES 128
+#define TX_RING_MASK (TX_RING_ENTRIES - 1)

#define ETCSR_FD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
#define ETCSR_HD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
@@ -76,8 +82,8 @@ struct ioc3_private {
u32 *ssram;
unsigned long *rxr; /* pointer to receiver ring */
struct ioc3_etxd *txr;
- struct sk_buff *rx_skbs[512];
- struct sk_buff *tx_skbs[128];
+ struct sk_buff *rx_skbs[RX_RING_ENTRIES];
+ struct sk_buff *tx_skbs[TX_RING_ENTRIES];
int rx_ci; /* RX consumer index */
int rx_pi; /* RX producer index */
int tx_ci; /* TX consumer index */
@@ -573,10 +579,10 @@ static inline void ioc3_rx(struct net_device *dev)
ip->rx_skbs[n_entry] = new_skb;
rxr[n_entry] = cpu_to_be64(ioc3_map(rxb, 1));
rxb->w0 = 0; /* Clear valid flag */
- n_entry = (n_entry + 1) & 511; /* Update erpir */
+ n_entry = (n_entry + 1) & RX_RING_MASK; /* Update erpir */

/* Now go on to the next ring entry. */
- rx_entry = (rx_entry + 1) & 511;
+ rx_entry = (rx_entry + 1) & RX_RING_MASK;
skb = ip->rx_skbs[rx_entry];
rxb = (struct ioc3_erxbuf *)(skb->data - RX_OFFSET);
w0 = be32_to_cpu(rxb->w0);
@@ -598,7 +604,7 @@ static inline void ioc3_tx(struct net_device *dev)
spin_lock(&ip->ioc3_lock);
etcir = readl(&regs->etcir);

- tx_entry = (etcir >> 7) & 127;
+ tx_entry = (etcir >> 7) & TX_RING_MASK;
o_entry = ip->tx_ci;
packets = 0;
bytes = 0;
@@ -610,17 +616,17 @@ static inline void ioc3_tx(struct net_device *dev)
dev_consume_skb_irq(skb);
ip->tx_skbs[o_entry] = NULL;

- o_entry = (o_entry + 1) & 127; /* Next */
+ o_entry = (o_entry + 1) & TX_RING_MASK; /* Next */

etcir = readl(&regs->etcir); /* More pkts sent? */
- tx_entry = (etcir >> 7) & 127;
+ tx_entry = (etcir >> 7) & TX_RING_MASK;
}

dev->stats.tx_packets += packets;
dev->stats.tx_bytes += bytes;
ip->txqlen -= packets;

- if (ip->txqlen < 128)
+ if (netif_queue_stopped(dev) && ip->txqlen < TX_RING_ENTRIES)
netif_wake_queue(dev);

ip->tx_ci = o_entry;
@@ -765,10 +771,10 @@ static inline void ioc3_clean_rx_ring(struct ioc3_private *ip)
ip->rx_skbs[ip->rx_pi] = ip->rx_skbs[ip->rx_ci];
ip->rxr[ip->rx_pi++] = ip->rxr[ip->rx_ci++];
}
- ip->rx_pi &= 511;
- ip->rx_ci &= 511;
+ ip->rx_pi &= RX_RING_MASK;
+ ip->rx_ci &= RX_RING_MASK;

- for (i = ip->rx_ci; i != ip->rx_pi; i = (i + 1) & 511) {
+ for (i = ip->rx_ci; i != ip->rx_pi; i = (i + 1) & RX_RING_MASK) {
skb = ip->rx_skbs[i];
rxb = (struct ioc3_erxbuf *)(skb->data - RX_OFFSET);
rxb->w0 = 0;
@@ -780,7 +786,7 @@ static inline void ioc3_clean_tx_ring(struct ioc3_private *ip)
struct sk_buff *skb;
int i;

- for (i = 0; i < 128; i++) {
+ for (i = 0; i < TX_RING_ENTRIES; i++) {
skb = ip->tx_skbs[i];
if (skb) {
ip->tx_skbs[i] = NULL;
@@ -812,7 +818,7 @@ static void ioc3_free_rings(struct ioc3_private *ip)
if (skb)
dev_kfree_skb_any(skb);

- n_entry = (n_entry + 1) & 511;
+ n_entry = (n_entry + 1) & RX_RING_MASK;
}
free_page((unsigned long)ip->rxr);
ip->rxr = NULL;
@@ -1425,13 +1431,13 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
mb(); /* make sure all descriptor changes are visible */

ip->tx_skbs[produce] = skb; /* Remember skb */
- produce = (produce + 1) & 127;
+ produce = (produce + 1) & TX_RING_MASK;
ip->tx_pi = produce;
writel(produce << 7, &ip->regs->etpir); /* Fire ... */

ip->txqlen++;

- if (ip->txqlen >= 127)
+ if (ip->txqlen >= (TX_RING_ENTRIES - 1))
netif_stop_queue(dev);

spin_unlock_irq(&ip->ioc3_lock);
--
2.13.7

2019-08-19 16:35:20

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 04/17] MIPS: PCI: refactor ioc3 special handling

Refactored code to only have one ioc3 special handling for read
access and one for write access.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
arch/mips/pci/pci-xtalk-bridge.c | 167 +++++++++++++++------------------------
1 file changed, 62 insertions(+), 105 deletions(-)

diff --git a/arch/mips/pci/pci-xtalk-bridge.c b/arch/mips/pci/pci-xtalk-bridge.c
index bcf7f559789a..7b4d40354ee7 100644
--- a/arch/mips/pci/pci-xtalk-bridge.c
+++ b/arch/mips/pci/pci-xtalk-bridge.c
@@ -20,16 +20,50 @@
* Most of the IOC3 PCI config register aren't present
* we emulate what is needed for a normal PCI enumeration
*/
-static u32 emulate_ioc3_cfg(int where, int size)
+static int ioc3_cfg_rd(void *addr, int where, int size, u32 *value)
{
- if (size == 1 && where == 0x3d)
- return 0x01;
- else if (size == 2 && where == 0x3c)
- return 0x0100;
- else if (size == 4 && where == 0x3c)
- return 0x00000100;
+ u32 cf, shift, mask;

- return 0;
+ switch (where & ~3) {
+ case 0x00 ... 0x10:
+ case 0x40 ... 0x44:
+ if (get_dbe(cf, (u32 *)addr))
+ return PCIBIOS_DEVICE_NOT_FOUND;
+ break;
+ case 0x3c:
+ /* emulate sane interrupt pin value */
+ cf = 0x00000100;
+ break;
+ default:
+ cf = 0;
+ break;
+ }
+ shift = (where & 3) << 3;
+ mask = 0xffffffffU >> ((4 - size) << 3);
+ *value = (cf >> shift) & mask;
+
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static int ioc3_cfg_wr(void *addr, int where, int size, u32 value)
+{
+ u32 cf, shift, mask, smask;
+
+ if ((where >= 0x14 && where < 0x40) || (where >= 0x48))
+ return PCIBIOS_SUCCESSFUL;
+
+ if (get_dbe(cf, (u32 *)addr))
+ return PCIBIOS_DEVICE_NOT_FOUND;
+
+ shift = ((where & 3) << 3);
+ mask = (0xffffffffU >> ((4 - size) << 3));
+ smask = mask << shift;
+
+ cf = (cf & ~smask) | ((value & mask) << shift);
+ if (put_dbe(cf, (u32 *)addr))
+ return PCIBIOS_DEVICE_NOT_FOUND;
+
+ return PCIBIOS_SUCCESSFUL;
}

static void bridge_disable_swapping(struct pci_dev *dev)
@@ -64,7 +98,7 @@ static int pci_conf0_read_config(struct pci_bus *bus, unsigned int devfn,
int slot = PCI_SLOT(devfn);
int fn = PCI_FUNC(devfn);
void *addr;
- u32 cf, shift, mask;
+ u32 cf;
int res;

addr = &bridge->b_type0_cfg_dev[slot].f[fn].c[PCI_VENDOR_ID];
@@ -75,8 +109,10 @@ static int pci_conf0_read_config(struct pci_bus *bus, unsigned int devfn,
* IOC3 is broken beyond belief ... Don't even give the
* generic PCI code a chance to look at it for real ...
*/
- if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16)))
- goto is_ioc3;
+ if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16))) {
+ addr = &bridge->b_type0_cfg_dev[slot].f[fn].l[where >> 2];
+ return ioc3_cfg_rd(addr, where, size, value);
+ }

addr = &bridge->b_type0_cfg_dev[slot].f[fn].c[where ^ (4 - size)];

@@ -88,26 +124,6 @@ static int pci_conf0_read_config(struct pci_bus *bus, unsigned int devfn,
res = get_dbe(*value, (u32 *)addr);

return res ? PCIBIOS_DEVICE_NOT_FOUND : PCIBIOS_SUCCESSFUL;
-
-is_ioc3:
-
- /*
- * IOC3 special handling
- */
- if ((where >= 0x14 && where < 0x40) || (where >= 0x48)) {
- *value = emulate_ioc3_cfg(where, size);
- return PCIBIOS_SUCCESSFUL;
- }
-
- addr = &bridge->b_type0_cfg_dev[slot].f[fn].l[where >> 2];
- if (get_dbe(cf, (u32 *)addr))
- return PCIBIOS_DEVICE_NOT_FOUND;
-
- shift = ((where & 3) << 3);
- mask = (0xffffffffU >> ((4 - size) << 3));
- *value = (cf >> shift) & mask;
-
- return PCIBIOS_SUCCESSFUL;
}

static int pci_conf1_read_config(struct pci_bus *bus, unsigned int devfn,
@@ -119,7 +135,7 @@ static int pci_conf1_read_config(struct pci_bus *bus, unsigned int devfn,
int slot = PCI_SLOT(devfn);
int fn = PCI_FUNC(devfn);
void *addr;
- u32 cf, shift, mask;
+ u32 cf;
int res;

bridge_write(bc, b_pci_cfg, (busno << 16) | (slot << 11));
@@ -131,8 +147,10 @@ static int pci_conf1_read_config(struct pci_bus *bus, unsigned int devfn,
* IOC3 is broken beyond belief ... Don't even give the
* generic PCI code a chance to look at it for real ...
*/
- if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16)))
- goto is_ioc3;
+ if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16))) {
+ addr = &bridge->b_type1_cfg.c[(fn << 8) | (where & ~3)];
+ return ioc3_cfg_rd(addr, where, size, value);
+ }

addr = &bridge->b_type1_cfg.c[(fn << 8) | (where ^ (4 - size))];

@@ -144,26 +162,6 @@ static int pci_conf1_read_config(struct pci_bus *bus, unsigned int devfn,
res = get_dbe(*value, (u32 *)addr);

return res ? PCIBIOS_DEVICE_NOT_FOUND : PCIBIOS_SUCCESSFUL;
-
-is_ioc3:
-
- /*
- * IOC3 special handling
- */
- if ((where >= 0x14 && where < 0x40) || (where >= 0x48)) {
- *value = emulate_ioc3_cfg(where, size);
- return PCIBIOS_SUCCESSFUL;
- }
-
- addr = &bridge->b_type1_cfg.c[(fn << 8) | where];
- if (get_dbe(cf, (u32 *)addr))
- return PCIBIOS_DEVICE_NOT_FOUND;
-
- shift = ((where & 3) << 3);
- mask = (0xffffffffU >> ((4 - size) << 3));
- *value = (cf >> shift) & mask;
-
- return PCIBIOS_SUCCESSFUL;
}

static int pci_read_config(struct pci_bus *bus, unsigned int devfn,
@@ -183,7 +181,7 @@ static int pci_conf0_write_config(struct pci_bus *bus, unsigned int devfn,
int slot = PCI_SLOT(devfn);
int fn = PCI_FUNC(devfn);
void *addr;
- u32 cf, shift, mask, smask;
+ u32 cf;
int res;

addr = &bridge->b_type0_cfg_dev[slot].f[fn].c[PCI_VENDOR_ID];
@@ -194,8 +192,10 @@ static int pci_conf0_write_config(struct pci_bus *bus, unsigned int devfn,
* IOC3 is broken beyond belief ... Don't even give the
* generic PCI code a chance to look at it for real ...
*/
- if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16)))
- goto is_ioc3;
+ if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16))) {
+ addr = &bridge->b_type0_cfg_dev[slot].f[fn].l[where >> 2];
+ return ioc3_cfg_wr(addr, where, size, value);
+ }

addr = &bridge->b_type0_cfg_dev[slot].f[fn].c[where ^ (4 - size)];

@@ -210,29 +210,6 @@ static int pci_conf0_write_config(struct pci_bus *bus, unsigned int devfn,
return PCIBIOS_DEVICE_NOT_FOUND;

return PCIBIOS_SUCCESSFUL;
-
-is_ioc3:
-
- /*
- * IOC3 special handling
- */
- if ((where >= 0x14 && where < 0x40) || (where >= 0x48))
- return PCIBIOS_SUCCESSFUL;
-
- addr = &bridge->b_type0_cfg_dev[slot].f[fn].l[where >> 2];
-
- if (get_dbe(cf, (u32 *)addr))
- return PCIBIOS_DEVICE_NOT_FOUND;
-
- shift = ((where & 3) << 3);
- mask = (0xffffffffU >> ((4 - size) << 3));
- smask = mask << shift;
-
- cf = (cf & ~smask) | ((value & mask) << shift);
- if (put_dbe(cf, (u32 *)addr))
- return PCIBIOS_DEVICE_NOT_FOUND;
-
- return PCIBIOS_SUCCESSFUL;
}

static int pci_conf1_write_config(struct pci_bus *bus, unsigned int devfn,
@@ -244,7 +221,7 @@ static int pci_conf1_write_config(struct pci_bus *bus, unsigned int devfn,
int fn = PCI_FUNC(devfn);
int busno = bus->number;
void *addr;
- u32 cf, shift, mask, smask;
+ u32 cf;
int res;

bridge_write(bc, b_pci_cfg, (busno << 16) | (slot << 11));
@@ -256,8 +233,10 @@ static int pci_conf1_write_config(struct pci_bus *bus, unsigned int devfn,
* IOC3 is broken beyond belief ... Don't even give the
* generic PCI code a chance to look at it for real ...
*/
- if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16)))
- goto is_ioc3;
+ if (cf == (PCI_VENDOR_ID_SGI | (PCI_DEVICE_ID_SGI_IOC3 << 16))) {
+ addr = &bridge->b_type0_cfg_dev[slot].f[fn].l[where >> 2];
+ return ioc3_cfg_wr(addr, where, size, value);
+ }

addr = &bridge->b_type1_cfg.c[(fn << 8) | (where ^ (4 - size))];

@@ -272,28 +251,6 @@ static int pci_conf1_write_config(struct pci_bus *bus, unsigned int devfn,
return PCIBIOS_DEVICE_NOT_FOUND;

return PCIBIOS_SUCCESSFUL;
-
-is_ioc3:
-
- /*
- * IOC3 special handling
- */
- if ((where >= 0x14 && where < 0x40) || (where >= 0x48))
- return PCIBIOS_SUCCESSFUL;
-
- addr = &bridge->b_type0_cfg_dev[slot].f[fn].l[where >> 2];
- if (get_dbe(cf, (u32 *)addr))
- return PCIBIOS_DEVICE_NOT_FOUND;
-
- shift = ((where & 3) << 3);
- mask = (0xffffffffU >> ((4 - size) << 3));
- smask = mask << shift;
-
- cf = (cf & ~smask) | ((value & mask) << shift);
- if (put_dbe(cf, (u32 *)addr))
- return PCIBIOS_DEVICE_NOT_FOUND;
-
- return PCIBIOS_SUCCESSFUL;
}

static int pci_write_config(struct pci_bus *bus, unsigned int devfn,
--
2.13.7

2019-08-19 16:35:29

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 01/17] w1: add 1-wire master driver for IP block found in SGI ASICs

Starting with SGI Origin machines nearly every new SGI ASIC contains
an 1-Wire master. They are used for attaching One-Wire prom devices,
which contain information about part numbers, revision numbers,
serial number etc. and MAC addresses for ethernet interfaces.
This patch adds a master driver to support this IP block.
It also adds an extra field dev_id to struct w1_bus_master, which
could be in used in slave drivers for creating unique device names.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/w1/masters/Kconfig | 9 +++
drivers/w1/masters/Makefile | 1 +
drivers/w1/masters/sgi_w1.c | 130 +++++++++++++++++++++++++++++++++++
include/linux/platform_data/sgi-w1.h | 15 ++++
include/linux/w1.h | 2 +
5 files changed, 157 insertions(+)
create mode 100644 drivers/w1/masters/sgi_w1.c
create mode 100644 include/linux/platform_data/sgi-w1.h

diff --git a/drivers/w1/masters/Kconfig b/drivers/w1/masters/Kconfig
index 7ae260577901..24b9a8e05f64 100644
--- a/drivers/w1/masters/Kconfig
+++ b/drivers/w1/masters/Kconfig
@@ -65,5 +65,14 @@ config HDQ_MASTER_OMAP
Say Y here if you want support for the 1-wire or HDQ Interface
on an OMAP processor.

+config W1_MASTER_SGI
+ tristate "SGI ASIC driver"
+ help
+ Say Y here if you want support for your 1-wire devices using
+ SGI ASIC 1-Wire interface
+
+ This support is also available as a module. If so, the module
+ will be called sgi_w1.
+
endmenu

diff --git a/drivers/w1/masters/Makefile b/drivers/w1/masters/Makefile
index 18954cae4256..dae629b7ab49 100644
--- a/drivers/w1/masters/Makefile
+++ b/drivers/w1/masters/Makefile
@@ -11,3 +11,4 @@ obj-$(CONFIG_W1_MASTER_MXC) += mxc_w1.o
obj-$(CONFIG_W1_MASTER_DS1WM) += ds1wm.o
obj-$(CONFIG_W1_MASTER_GPIO) += w1-gpio.o
obj-$(CONFIG_HDQ_MASTER_OMAP) += omap_hdq.o
+obj-$(CONFIG_W1_MASTER_SGI) += sgi_w1.o
diff --git a/drivers/w1/masters/sgi_w1.c b/drivers/w1/masters/sgi_w1.c
new file mode 100644
index 000000000000..1b2d96b945be
--- /dev/null
+++ b/drivers/w1/masters/sgi_w1.c
@@ -0,0 +1,130 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * sgi_w1.c - w1 master driver for one wire support in SGI ASICs
+ */
+
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/io.h>
+#include <linux/jiffies.h>
+#include <linux/module.h>
+#include <linux/mod_devicetable.h>
+#include <linux/platform_device.h>
+#include <linux/platform_data/sgi-w1.h>
+
+#include <linux/w1.h>
+
+#define MCR_RD_DATA BIT(0)
+#define MCR_DONE BIT(1)
+
+#define MCR_PACK(pulse, sample) (((pulse) << 10) | ((sample) << 2))
+
+struct sgi_w1_device {
+ u32 __iomem *mcr;
+ struct w1_bus_master bus_master;
+ char dev_id[64];
+};
+
+static u8 sgi_w1_wait(u32 __iomem *mcr)
+{
+ u32 mcr_val;
+
+ do {
+ mcr_val = readl(mcr);
+ } while (!(mcr_val & MCR_DONE));
+
+ return (mcr_val & MCR_RD_DATA) ? 1 : 0;
+}
+
+/*
+ * this is the low level routine to
+ * reset the device on the One Wire interface
+ * on the hardware
+ */
+static u8 sgi_w1_reset_bus(void *data)
+{
+ struct sgi_w1_device *dev = data;
+ u8 ret;
+
+ writel(MCR_PACK(520, 65), dev->mcr);
+ ret = sgi_w1_wait(dev->mcr);
+ udelay(500); /* recovery time */
+ return ret;
+}
+
+/*
+ * this is the low level routine to read/write a bit on the One Wire
+ * interface on the hardware. It does write 0 if parameter bit is set
+ * to 0, otherwise a write 1/read.
+ */
+static u8 sgi_w1_touch_bit(void *data, u8 bit)
+{
+ struct sgi_w1_device *dev = data;
+ u8 ret;
+
+ if (bit)
+ writel(MCR_PACK(6, 13), dev->mcr);
+ else
+ writel(MCR_PACK(80, 30), dev->mcr);
+
+ ret = sgi_w1_wait(dev->mcr);
+ if (bit)
+ udelay(100); /* recovery */
+ return ret;
+}
+
+static int sgi_w1_probe(struct platform_device *pdev)
+{
+ struct sgi_w1_device *sdev;
+ struct sgi_w1_platform_data *pdata;
+ struct resource *res;
+
+ sdev = devm_kzalloc(&pdev->dev, sizeof(struct sgi_w1_device),
+ GFP_KERNEL);
+ if (!sdev)
+ return -ENOMEM;
+
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ sdev->mcr = devm_ioremap_resource(&pdev->dev, res);
+ if (IS_ERR(sdev->mcr))
+ return PTR_ERR(sdev->mcr);
+
+ sdev->bus_master.data = sdev;
+ sdev->bus_master.reset_bus = sgi_w1_reset_bus;
+ sdev->bus_master.touch_bit = sgi_w1_touch_bit;
+
+ pdata = dev_get_platdata(&pdev->dev);
+ if (pdata) {
+ strlcpy(sdev->dev_id, pdata->dev_id, sizeof(sdev->dev_id));
+ sdev->bus_master.dev_id = sdev->dev_id;
+ }
+
+ platform_set_drvdata(pdev, sdev);
+
+ return w1_add_master_device(&sdev->bus_master);
+}
+
+/*
+ * disassociate the w1 device from the driver
+ */
+static int sgi_w1_remove(struct platform_device *pdev)
+{
+ struct sgi_w1_device *sdev = platform_get_drvdata(pdev);
+
+ w1_remove_master_device(&sdev->bus_master);
+
+ return 0;
+}
+
+static struct platform_driver sgi_w1_driver = {
+ .driver = {
+ .name = "sgi_w1",
+ },
+ .probe = sgi_w1_probe,
+ .remove = sgi_w1_remove,
+};
+module_platform_driver(sgi_w1_driver);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Thomas Bogendoerfer");
+MODULE_DESCRIPTION("Driver for One-Wire IP in SGI ASICs");
diff --git a/include/linux/platform_data/sgi-w1.h b/include/linux/platform_data/sgi-w1.h
new file mode 100644
index 000000000000..fc6b92e0b942
--- /dev/null
+++ b/include/linux/platform_data/sgi-w1.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * SGI One-Wire (W1) IP
+ */
+
+#ifndef PLATFORM_DATA_SGI_W1_H
+#define PLATFORM_DATA_SGI_W1_H
+
+#include <asm/sn/types.h>
+
+struct sgi_w1_platform_data {
+ char dev_id[64];
+};
+
+#endif /* PLATFORM_DATA_SGI_W1_H */
diff --git a/include/linux/w1.h b/include/linux/w1.h
index e0b5156f78fd..89843e9f634c 100644
--- a/include/linux/w1.h
+++ b/include/linux/w1.h
@@ -150,6 +150,8 @@ struct w1_bus_master {

void (*search)(void *, struct w1_master *,
u8, w1_slave_found_callback);
+
+ char *dev_id;
};

/**
--
2.13.7

2019-08-19 16:35:48

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 14/17] net: sgi: ioc3-eth: Fix IPG settings

The half/full duplex settings for inter packet gap counters/timer were
reversed.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 56392f0b01de..25597c6194d9 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -80,8 +80,8 @@
#define RX_OFFSET 10
#define RX_BUF_SIZE 1664

-#define ETCSR_FD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
-#define ETCSR_HD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
+#define ETCSR_FD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
+#define ETCSR_HD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)

/* Private per NIC data of the driver. */
struct ioc3_private {
--
2.13.7

2019-08-19 16:36:10

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 07/17] MIPS: SGI-IP27: restructure ioc3 register access

Break up the big ioc3 register struct into functional pieces to
make use in sub-function drivers more straightforward. And while
doing that get rid of all volatile access by using readX/writeX.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
arch/mips/include/asm/sn/ioc3.h | 359 +++++++++++++------------------
arch/mips/sgi-ip27/ip27-console.c | 5 +-
drivers/net/ethernet/sgi/ioc3-eth.c | 418 ++++++++++++++++--------------------
3 files changed, 334 insertions(+), 448 deletions(-)

diff --git a/arch/mips/include/asm/sn/ioc3.h b/arch/mips/include/asm/sn/ioc3.h
index 5022b0ab2074..8ed952d93252 100644
--- a/arch/mips/include/asm/sn/ioc3.h
+++ b/arch/mips/include/asm/sn/ioc3.h
@@ -3,169 +3,161 @@
* Copyright (C) 1999, 2000 Ralf Baechle
* Copyright (C) 1999, 2000 Silicon Graphics, Inc.
*/
-#ifndef _IOC3_H
-#define _IOC3_H
+#ifndef MIPS_SN_IOC3_H
+#define MIPS_SN_IOC3_H

#include <linux/types.h>

+/* serial port register map */
+struct ioc3_serialregs {
+ u32 sscr;
+ u32 stpir;
+ u32 stcir;
+ u32 srpir;
+ u32 srcir;
+ u32 srtr;
+ u32 shadow;
+};
+
/* SUPERIO uart register map */
-typedef volatile struct ioc3_uartregs {
+struct ioc3_uartregs {
union {
- volatile u8 rbr; /* read only, DLAB == 0 */
- volatile u8 thr; /* write only, DLAB == 0 */
- volatile u8 dll; /* DLAB == 1 */
- } u1;
+ u8 iu_rbr; /* read only, DLAB == 0 */
+ u8 iu_thr; /* write only, DLAB == 0 */
+ u8 iu_dll; /* DLAB == 1 */
+ };
union {
- volatile u8 ier; /* DLAB == 0 */
- volatile u8 dlm; /* DLAB == 1 */
- } u2;
+ u8 iu_ier; /* DLAB == 0 */
+ u8 iu_dlm; /* DLAB == 1 */
+ };
union {
- volatile u8 iir; /* read only */
- volatile u8 fcr; /* write only */
- } u3;
- volatile u8 iu_lcr;
- volatile u8 iu_mcr;
- volatile u8 iu_lsr;
- volatile u8 iu_msr;
- volatile u8 iu_scr;
-} ioc3_uregs_t;
-
-#define iu_rbr u1.rbr
-#define iu_thr u1.thr
-#define iu_dll u1.dll
-#define iu_ier u2.ier
-#define iu_dlm u2.dlm
-#define iu_iir u3.iir
-#define iu_fcr u3.fcr
+ u8 iu_iir; /* read only */
+ u8 iu_fcr; /* write only */
+ };
+ u8 iu_lcr;
+ u8 iu_mcr;
+ u8 iu_lsr;
+ u8 iu_msr;
+ u8 iu_scr;
+};

struct ioc3_sioregs {
- volatile u8 fill[0x141]; /* starts at 0x141 */
+ u8 fill[0x141]; /* starts at 0x141 */

- volatile u8 uartc;
- volatile u8 kbdcg;
+ u8 uartc;
+ u8 kbdcg;

- volatile u8 fill0[0x150 - 0x142 - 1];
+ u8 fill0[0x150 - 0x142 - 1];

- volatile u8 pp_data;
- volatile u8 pp_dsr;
- volatile u8 pp_dcr;
+ u8 pp_data;
+ u8 pp_dsr;
+ u8 pp_dcr;

- volatile u8 fill1[0x158 - 0x152 - 1];
+ u8 fill1[0x158 - 0x152 - 1];

- volatile u8 pp_fifa;
- volatile u8 pp_cfgb;
- volatile u8 pp_ecr;
+ u8 pp_fifa;
+ u8 pp_cfgb;
+ u8 pp_ecr;

- volatile u8 fill2[0x168 - 0x15a - 1];
+ u8 fill2[0x168 - 0x15a - 1];

- volatile u8 rtcad;
- volatile u8 rtcdat;
+ u8 rtcad;
+ u8 rtcdat;

- volatile u8 fill3[0x170 - 0x169 - 1];
+ u8 fill3[0x170 - 0x169 - 1];

struct ioc3_uartregs uartb; /* 0x20170 */
struct ioc3_uartregs uarta; /* 0x20178 */
};

+struct ioc3_ethregs {
+ u32 emcr; /* 0x000f0 */
+ u32 eisr; /* 0x000f4 */
+ u32 eier; /* 0x000f8 */
+ u32 ercsr; /* 0x000fc */
+ u32 erbr_h; /* 0x00100 */
+ u32 erbr_l; /* 0x00104 */
+ u32 erbar; /* 0x00108 */
+ u32 ercir; /* 0x0010c */
+ u32 erpir; /* 0x00110 */
+ u32 ertr; /* 0x00114 */
+ u32 etcsr; /* 0x00118 */
+ u32 ersr; /* 0x0011c */
+ u32 etcdc; /* 0x00120 */
+ u32 ebir; /* 0x00124 */
+ u32 etbr_h; /* 0x00128 */
+ u32 etbr_l; /* 0x0012c */
+ u32 etcir; /* 0x00130 */
+ u32 etpir; /* 0x00134 */
+ u32 emar_h; /* 0x00138 */
+ u32 emar_l; /* 0x0013c */
+ u32 ehar_h; /* 0x00140 */
+ u32 ehar_l; /* 0x00144 */
+ u32 micr; /* 0x00148 */
+ u32 midr_r; /* 0x0014c */
+ u32 midr_w; /* 0x00150 */
+};
+
+struct ioc3_serioregs {
+ u32 km_csr; /* 0x0009c */
+ u32 k_rd; /* 0x000a0 */
+ u32 m_rd; /* 0x000a4 */
+ u32 k_wd; /* 0x000a8 */
+ u32 m_wd; /* 0x000ac */
+};
+
/* Register layout of IOC3 in configuration space. */
struct ioc3 {
- volatile u32 pad0[7]; /* 0x00000 */
- volatile u32 sio_ir; /* 0x0001c */
- volatile u32 sio_ies; /* 0x00020 */
- volatile u32 sio_iec; /* 0x00024 */
- volatile u32 sio_cr; /* 0x00028 */
- volatile u32 int_out; /* 0x0002c */
- volatile u32 mcr; /* 0x00030 */
+ /* PCI Config Space registers */
+ u32 pci_id; /* 0x00000 */
+ u32 pci_scr; /* 0x00004 */
+ u32 pci_rev; /* 0x00008 */
+ u32 pci_lat; /* 0x0000c */
+ u32 pci_addr; /* 0x00010 */
+ u32 pci_err_addr_l; /* 0x00014 */
+ u32 pci_err_addr_h; /* 0x00018 */
+
+ u32 sio_ir; /* 0x0001c */
+ u32 sio_ies; /* 0x00020 */
+ u32 sio_iec; /* 0x00024 */
+ u32 sio_cr; /* 0x00028 */
+ u32 int_out; /* 0x0002c */
+ u32 mcr; /* 0x00030 */

/* General Purpose I/O registers */
- volatile u32 gpcr_s; /* 0x00034 */
- volatile u32 gpcr_c; /* 0x00038 */
- volatile u32 gpdr; /* 0x0003c */
- volatile u32 gppr_0; /* 0x00040 */
- volatile u32 gppr_1; /* 0x00044 */
- volatile u32 gppr_2; /* 0x00048 */
- volatile u32 gppr_3; /* 0x0004c */
- volatile u32 gppr_4; /* 0x00050 */
- volatile u32 gppr_5; /* 0x00054 */
- volatile u32 gppr_6; /* 0x00058 */
- volatile u32 gppr_7; /* 0x0005c */
- volatile u32 gppr_8; /* 0x00060 */
- volatile u32 gppr_9; /* 0x00064 */
- volatile u32 gppr_10; /* 0x00068 */
- volatile u32 gppr_11; /* 0x0006c */
- volatile u32 gppr_12; /* 0x00070 */
- volatile u32 gppr_13; /* 0x00074 */
- volatile u32 gppr_14; /* 0x00078 */
- volatile u32 gppr_15; /* 0x0007c */
+ u32 gpcr_s; /* 0x00034 */
+ u32 gpcr_c; /* 0x00038 */
+ u32 gpdr; /* 0x0003c */
+ u32 gppr[16]; /* 0x00040 */

/* Parallel Port Registers */
- volatile u32 ppbr_h_a; /* 0x00080 */
- volatile u32 ppbr_l_a; /* 0x00084 */
- volatile u32 ppcr_a; /* 0x00088 */
- volatile u32 ppcr; /* 0x0008c */
- volatile u32 ppbr_h_b; /* 0x00090 */
- volatile u32 ppbr_l_b; /* 0x00094 */
- volatile u32 ppcr_b; /* 0x00098 */
+ u32 ppbr_h_a; /* 0x00080 */
+ u32 ppbr_l_a; /* 0x00084 */
+ u32 ppcr_a; /* 0x00088 */
+ u32 ppcr; /* 0x0008c */
+ u32 ppbr_h_b; /* 0x00090 */
+ u32 ppbr_l_b; /* 0x00094 */
+ u32 ppcr_b; /* 0x00098 */

/* Keyboard and Mouse Registers */
- volatile u32 km_csr; /* 0x0009c */
- volatile u32 k_rd; /* 0x000a0 */
- volatile u32 m_rd; /* 0x000a4 */
- volatile u32 k_wd; /* 0x000a8 */
- volatile u32 m_wd; /* 0x000ac */
+ struct ioc3_serioregs serio;

/* Serial Port Registers */
- volatile u32 sbbr_h; /* 0x000b0 */
- volatile u32 sbbr_l; /* 0x000b4 */
- volatile u32 sscr_a; /* 0x000b8 */
- volatile u32 stpir_a; /* 0x000bc */
- volatile u32 stcir_a; /* 0x000c0 */
- volatile u32 srpir_a; /* 0x000c4 */
- volatile u32 srcir_a; /* 0x000c8 */
- volatile u32 srtr_a; /* 0x000cc */
- volatile u32 shadow_a; /* 0x000d0 */
- volatile u32 sscr_b; /* 0x000d4 */
- volatile u32 stpir_b; /* 0x000d8 */
- volatile u32 stcir_b; /* 0x000dc */
- volatile u32 srpir_b; /* 0x000e0 */
- volatile u32 srcir_b; /* 0x000e4 */
- volatile u32 srtr_b; /* 0x000e8 */
- volatile u32 shadow_b; /* 0x000ec */
-
- /* Ethernet Registers */
- volatile u32 emcr; /* 0x000f0 */
- volatile u32 eisr; /* 0x000f4 */
- volatile u32 eier; /* 0x000f8 */
- volatile u32 ercsr; /* 0x000fc */
- volatile u32 erbr_h; /* 0x00100 */
- volatile u32 erbr_l; /* 0x00104 */
- volatile u32 erbar; /* 0x00108 */
- volatile u32 ercir; /* 0x0010c */
- volatile u32 erpir; /* 0x00110 */
- volatile u32 ertr; /* 0x00114 */
- volatile u32 etcsr; /* 0x00118 */
- volatile u32 ersr; /* 0x0011c */
- volatile u32 etcdc; /* 0x00120 */
- volatile u32 ebir; /* 0x00124 */
- volatile u32 etbr_h; /* 0x00128 */
- volatile u32 etbr_l; /* 0x0012c */
- volatile u32 etcir; /* 0x00130 */
- volatile u32 etpir; /* 0x00134 */
- volatile u32 emar_h; /* 0x00138 */
- volatile u32 emar_l; /* 0x0013c */
- volatile u32 ehar_h; /* 0x00140 */
- volatile u32 ehar_l; /* 0x00144 */
- volatile u32 micr; /* 0x00148 */
- volatile u32 midr_r; /* 0x0014c */
- volatile u32 midr_w; /* 0x00150 */
- volatile u32 pad1[(0x20000 - 0x00154) / 4];
+ u32 sbbr_h; /* 0x000b0 */
+ u32 sbbr_l; /* 0x000b4 */
+ struct ioc3_serialregs port_a;
+ struct ioc3_serialregs port_b;
+
+ /* Ethernet Registers */
+ struct ioc3_ethregs eth;
+ u32 pad1[(0x20000 - 0x00154) / 4];

/* SuperIO Registers XXX */
struct ioc3_sioregs sregs; /* 0x20000 */
- volatile u32 pad2[(0x40000 - 0x20180) / 4];
+ u32 pad2[(0x40000 - 0x20180) / 4];

/* SSRAM Diagnostic Access */
- volatile u32 ssram[(0x80000 - 0x40000) / 4];
+ u32 ssram[(0x80000 - 0x40000) / 4];

/* Bytebus device offsets
0x80000 - Access to the generic devices selected with DEV0
@@ -178,6 +170,20 @@ struct ioc3 {
0xFFFFF bytebus DEV_SEL_3 */
};

+
+#define PCI_LAT 0xc /* Latency Timer */
+#define PCI_SCR_DROP_MODE_EN 0x00008000 /* drop pios on parity err */
+#define UARTA_BASE 0x178
+#define UARTB_BASE 0x170
+
+/*
+ * Bytebus device space
+ */
+#define IOC3_BYTEBUS_DEV0 0x80000L
+#define IOC3_BYTEBUS_DEV1 0xa0000L
+#define IOC3_BYTEBUS_DEV2 0xc0000L
+#define IOC3_BYTEBUS_DEV3 0xe0000L
+
/*
* Ethernet RX Buffer
*/
@@ -233,28 +239,20 @@ struct ioc3_etxd {
#define ETXD_B2CNT_MASK 0x7ff00000
#define ETXD_B2CNT_SHIFT 20

-/*
- * Bytebus device space
- */
-#define IOC3_BYTEBUS_DEV0 0x80000L
-#define IOC3_BYTEBUS_DEV1 0xa0000L
-#define IOC3_BYTEBUS_DEV2 0xc0000L
-#define IOC3_BYTEBUS_DEV3 0xe0000L
-
/* ------------------------------------------------------------------------- */

/* Superio Registers (PIO Access) */
#define IOC3_SIO_BASE 0x20000
#define IOC3_SIO_UARTC (IOC3_SIO_BASE+0x141) /* UART Config */
#define IOC3_SIO_KBDCG (IOC3_SIO_BASE+0x142) /* KBD Config */
-#define IOC3_SIO_PP_BASE (IOC3_SIO_BASE+PP_BASE) /* Parallel Port */
+#define IOC3_SIO_PP_BASE (IOC3_SIO_BASE+PP_BASE) /* Parallel Port */
#define IOC3_SIO_RTC_BASE (IOC3_SIO_BASE+0x168) /* Real Time Clock */
#define IOC3_SIO_UB_BASE (IOC3_SIO_BASE+UARTB_BASE) /* UART B */
#define IOC3_SIO_UA_BASE (IOC3_SIO_BASE+UARTA_BASE) /* UART A */

/* SSRAM Diagnostic Access */
#define IOC3_SSRAM IOC3_RAM_OFF /* base of SSRAM diagnostic access */
-#define IOC3_SSRAM_LEN 0x40000 /* 256kb (address space size, may not be fully populated) */
+#define IOC3_SSRAM_LEN 0x40000 /* 256kb (addrspc sz, may not be populated) */
#define IOC3_SSRAM_DM 0x0000ffff /* data mask */
#define IOC3_SSRAM_PM 0x00010000 /* parity mask */

@@ -294,10 +292,10 @@ struct ioc3_etxd {
SIO_IR to assert */
#define KM_CSR_M_TO_EN 0x00080000 /* KM_CSR_M_TO + KM_CSR_M_TO_EN = cause
SIO_IR to assert */
-#define KM_CSR_K_CLAMP_ONE 0x00100000 /* Pull K_CLK low after rec. one char */
-#define KM_CSR_M_CLAMP_ONE 0x00200000 /* Pull M_CLK low after rec. one char */
-#define KM_CSR_K_CLAMP_THREE 0x00400000 /* Pull K_CLK low after rec. three chars */
-#define KM_CSR_M_CLAMP_THREE 0x00800000 /* Pull M_CLK low after rec. three char */
+#define KM_CSR_K_CLAMP_1 0x00100000 /* Pull K_CLK low aft recv 1 char */
+#define KM_CSR_M_CLAMP_1 0x00200000 /* Pull M_CLK low aft recv 1 char */
+#define KM_CSR_K_CLAMP_3 0x00400000 /* Pull K_CLK low aft recv 3 chars */
+#define KM_CSR_M_CLAMP_3 0x00800000 /* Pull M_CLK low aft recv 3 chars */

/* bitmasks for IOC3_K_RD and IOC3_M_RD */
#define KM_RD_DATA_2 0x000000ff /* 3rd char recvd since last read */
@@ -440,10 +438,6 @@ struct ioc3_etxd {
SIO_IR_PP_INTB | SIO_IR_PP_MEMERR)
#define SIO_IR_RT (SIO_IR_RT_INT | SIO_IR_GEN_INT1)

-/* macro to load pending interrupts */
-#define IOC3_PENDING_INTRS(mem) (PCI_INW(&((mem)->sio_ir)) & \
- PCI_INW(&((mem)->sio_ies_ro)))
-
/* bitmasks for SIO_CR */
#define SIO_CR_SIO_RESET 0x00000001 /* reset the SIO */
#define SIO_CR_SER_A_BASE 0x000000fe /* DMA poll addr port A */
@@ -500,10 +494,11 @@ struct ioc3_etxd {
#define GPCR_UARTB_MODESEL 0x40 /* pin is output to port B mode sel */
#define GPCR_UARTA_MODESEL 0x80 /* pin is output to port A mode sel */

-#define GPPR_PHY_RESET_PIN 5 /* GIO pin controlling phy reset */
-#define GPPR_UARTB_MODESEL_PIN 6 /* GIO pin controlling uart b mode select */
-#define GPPR_UARTA_MODESEL_PIN 7 /* GIO pin controlling uart a mode select */
+#define GPPR_PHY_RESET_PIN 5 /* GIO pin cntrlling phy reset */
+#define GPPR_UARTB_MODESEL_PIN 6 /* GIO pin cntrlling uart b mode sel */
+#define GPPR_UARTA_MODESEL_PIN 7 /* GIO pin cntrlling uart a mode sel */

+/* ethernet */
#define EMCR_DUPLEX 0x00000001
#define EMCR_PROMISC 0x00000002
#define EMCR_PADEN 0x00000004
@@ -595,71 +590,9 @@ struct ioc3_etxd {

#define MIDR_DATA_MASK 0x0000ffff

-#define ERXBUF_IPCKSUM_MASK 0x0000ffff
-#define ERXBUF_BYTECNT_MASK 0x07ff0000
-#define ERXBUF_BYTECNT_SHIFT 16
-#define ERXBUF_V 0x80000000
-
-#define ERXBUF_CRCERR 0x00000001 /* aka RSV15 */
-#define ERXBUF_FRAMERR 0x00000002 /* aka RSV14 */
-#define ERXBUF_CODERR 0x00000004 /* aka RSV13 */
-#define ERXBUF_INVPREAMB 0x00000008 /* aka RSV18 */
-#define ERXBUF_LOLEN 0x00007000 /* aka RSV2_0 */
-#define ERXBUF_HILEN 0x03ff0000 /* aka RSV12_3 */
-#define ERXBUF_MULTICAST 0x04000000 /* aka RSV16 */
-#define ERXBUF_BROADCAST 0x08000000 /* aka RSV17 */
-#define ERXBUF_LONGEVENT 0x10000000 /* aka RSV19 */
-#define ERXBUF_BADPKT 0x20000000 /* aka RSV20 */
-#define ERXBUF_GOODPKT 0x40000000 /* aka RSV21 */
-#define ERXBUF_CARRIER 0x80000000 /* aka RSV22 */
-
-#define ETXD_BYTECNT_MASK 0x000007ff /* total byte count */
-#define ETXD_INTWHENDONE 0x00001000 /* intr when done */
-#define ETXD_D0V 0x00010000 /* data 0 valid */
-#define ETXD_B1V 0x00020000 /* buf 1 valid */
-#define ETXD_B2V 0x00040000 /* buf 2 valid */
-#define ETXD_DOCHECKSUM 0x00080000 /* insert ip cksum */
-#define ETXD_CHKOFF_MASK 0x07f00000 /* cksum byte offset */
-#define ETXD_CHKOFF_SHIFT 20
-
-#define ETXD_D0CNT_MASK 0x0000007f
-#define ETXD_B1CNT_MASK 0x0007ff00
-#define ETXD_B1CNT_SHIFT 8
-#define ETXD_B2CNT_MASK 0x7ff00000
-#define ETXD_B2CNT_SHIFT 20
-
-typedef enum ioc3_subdevs_e {
- ioc3_subdev_ether,
- ioc3_subdev_generic,
- ioc3_subdev_nic,
- ioc3_subdev_kbms,
- ioc3_subdev_ttya,
- ioc3_subdev_ttyb,
- ioc3_subdev_ecpp,
- ioc3_subdev_rt,
- ioc3_nsubdevs
-} ioc3_subdev_t;
-
-/* subdevice disable bits,
- * from the standard INFO_LBL_SUBDEVS
- */
-#define IOC3_SDB_ETHER (1<<ioc3_subdev_ether)
-#define IOC3_SDB_GENERIC (1<<ioc3_subdev_generic)
-#define IOC3_SDB_NIC (1<<ioc3_subdev_nic)
-#define IOC3_SDB_KBMS (1<<ioc3_subdev_kbms)
-#define IOC3_SDB_TTYA (1<<ioc3_subdev_ttya)
-#define IOC3_SDB_TTYB (1<<ioc3_subdev_ttyb)
-#define IOC3_SDB_ECPP (1<<ioc3_subdev_ecpp)
-#define IOC3_SDB_RT (1<<ioc3_subdev_rt)
-
-#define IOC3_ALL_SUBDEVS ((1<<ioc3_nsubdevs)-1)
-
-#define IOC3_SDB_SERIAL (IOC3_SDB_TTYA|IOC3_SDB_TTYB)
-
-#define IOC3_STD_SUBDEVS IOC3_ALL_SUBDEVS
-
-#define IOC3_INTA_SUBDEVS IOC3_SDB_ETHER
-#define IOC3_INTB_SUBDEVS (IOC3_SDB_GENERIC|IOC3_SDB_KBMS|IOC3_SDB_SERIAL|IOC3_SDB_ECPP|IOC3_SDB_RT)
+#if defined(CONFIG_SGI_IP27) || defined(CONFIG_SGI_IP30)
+extern int bridge_alloc_irq(struct pci_dev *dev);
+#endif

/* subsystem IDs supplied by card detection in pci-xtalk-bridge */
#define IOC3_SUBSYS_IP27_BASEIO6G 0xc300
@@ -670,4 +603,4 @@ typedef enum ioc3_subdevs_e {
#define IOC3_SUBSYS_MENET 0xc305
#define IOC3_SUBSYS_MENET4 0xc306

-#endif /* _IOC3_H */
+#endif /* MIPS_SN_IOC3_H */
diff --git a/arch/mips/sgi-ip27/ip27-console.c b/arch/mips/sgi-ip27/ip27-console.c
index 6bdb48d41276..5886bee89d06 100644
--- a/arch/mips/sgi-ip27/ip27-console.c
+++ b/arch/mips/sgi-ip27/ip27-console.c
@@ -35,6 +35,7 @@ void prom_putchar(char c)
{
struct ioc3_uartregs *uart = console_uart();

- while ((uart->iu_lsr & 0x20) == 0);
- uart->iu_thr = c;
+ while ((readb(&uart->iu_lsr) & 0x20) == 0)
+ ;
+ writeb(c, &uart->iu_thr);
}
diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index 358e66b81926..713d2472cb97 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -76,7 +76,9 @@

/* Private per NIC data of the driver. */
struct ioc3_private {
- struct ioc3 *regs;
+ struct ioc3_ethregs *regs;
+ struct ioc3 *all_regs;
+ u32 *ssram;
unsigned long *rxr; /* pointer to receiver ring */
struct ioc3_etxd *txr;
struct sk_buff *rx_skbs[512];
@@ -156,128 +158,67 @@ static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)

#define IOC3_SIZE 0x100000

-/*
- * IOC3 is a big endian device
- *
- * Unorthodox but makes the users of these macros more readable - the pointer
- * to the IOC3's memory mapped registers is expected as struct ioc3 * ioc3
- * in the environment.
- */
-#define ioc3_r_mcr() be32_to_cpu(ioc3->mcr)
-#define ioc3_w_mcr(v) do { ioc3->mcr = cpu_to_be32(v); } while (0)
-#define ioc3_w_gpcr_s(v) do { ioc3->gpcr_s = cpu_to_be32(v); } while (0)
-#define ioc3_r_emcr() be32_to_cpu(ioc3->emcr)
-#define ioc3_w_emcr(v) do { ioc3->emcr = cpu_to_be32(v); } while (0)
-#define ioc3_r_eisr() be32_to_cpu(ioc3->eisr)
-#define ioc3_w_eisr(v) do { ioc3->eisr = cpu_to_be32(v); } while (0)
-#define ioc3_r_eier() be32_to_cpu(ioc3->eier)
-#define ioc3_w_eier(v) do { ioc3->eier = cpu_to_be32(v); } while (0)
-#define ioc3_r_ercsr() be32_to_cpu(ioc3->ercsr)
-#define ioc3_w_ercsr(v) do { ioc3->ercsr = cpu_to_be32(v); } while (0)
-#define ioc3_r_erbr_h() be32_to_cpu(ioc3->erbr_h)
-#define ioc3_w_erbr_h(v) do { ioc3->erbr_h = cpu_to_be32(v); } while (0)
-#define ioc3_r_erbr_l() be32_to_cpu(ioc3->erbr_l)
-#define ioc3_w_erbr_l(v) do { ioc3->erbr_l = cpu_to_be32(v); } while (0)
-#define ioc3_r_erbar() be32_to_cpu(ioc3->erbar)
-#define ioc3_w_erbar(v) do { ioc3->erbar = cpu_to_be32(v); } while (0)
-#define ioc3_r_ercir() be32_to_cpu(ioc3->ercir)
-#define ioc3_w_ercir(v) do { ioc3->ercir = cpu_to_be32(v); } while (0)
-#define ioc3_r_erpir() be32_to_cpu(ioc3->erpir)
-#define ioc3_w_erpir(v) do { ioc3->erpir = cpu_to_be32(v); } while (0)
-#define ioc3_r_ertr() be32_to_cpu(ioc3->ertr)
-#define ioc3_w_ertr(v) do { ioc3->ertr = cpu_to_be32(v); } while (0)
-#define ioc3_r_etcsr() be32_to_cpu(ioc3->etcsr)
-#define ioc3_w_etcsr(v) do { ioc3->etcsr = cpu_to_be32(v); } while (0)
-#define ioc3_r_ersr() be32_to_cpu(ioc3->ersr)
-#define ioc3_w_ersr(v) do { ioc3->ersr = cpu_to_be32(v); } while (0)
-#define ioc3_r_etcdc() be32_to_cpu(ioc3->etcdc)
-#define ioc3_w_etcdc(v) do { ioc3->etcdc = cpu_to_be32(v); } while (0)
-#define ioc3_r_ebir() be32_to_cpu(ioc3->ebir)
-#define ioc3_w_ebir(v) do { ioc3->ebir = cpu_to_be32(v); } while (0)
-#define ioc3_r_etbr_h() be32_to_cpu(ioc3->etbr_h)
-#define ioc3_w_etbr_h(v) do { ioc3->etbr_h = cpu_to_be32(v); } while (0)
-#define ioc3_r_etbr_l() be32_to_cpu(ioc3->etbr_l)
-#define ioc3_w_etbr_l(v) do { ioc3->etbr_l = cpu_to_be32(v); } while (0)
-#define ioc3_r_etcir() be32_to_cpu(ioc3->etcir)
-#define ioc3_w_etcir(v) do { ioc3->etcir = cpu_to_be32(v); } while (0)
-#define ioc3_r_etpir() be32_to_cpu(ioc3->etpir)
-#define ioc3_w_etpir(v) do { ioc3->etpir = cpu_to_be32(v); } while (0)
-#define ioc3_r_emar_h() be32_to_cpu(ioc3->emar_h)
-#define ioc3_w_emar_h(v) do { ioc3->emar_h = cpu_to_be32(v); } while (0)
-#define ioc3_r_emar_l() be32_to_cpu(ioc3->emar_l)
-#define ioc3_w_emar_l(v) do { ioc3->emar_l = cpu_to_be32(v); } while (0)
-#define ioc3_r_ehar_h() be32_to_cpu(ioc3->ehar_h)
-#define ioc3_w_ehar_h(v) do { ioc3->ehar_h = cpu_to_be32(v); } while (0)
-#define ioc3_r_ehar_l() be32_to_cpu(ioc3->ehar_l)
-#define ioc3_w_ehar_l(v) do { ioc3->ehar_l = cpu_to_be32(v); } while (0)
-#define ioc3_r_micr() be32_to_cpu(ioc3->micr)
-#define ioc3_w_micr(v) do { ioc3->micr = cpu_to_be32(v); } while (0)
-#define ioc3_r_midr_r() be32_to_cpu(ioc3->midr_r)
-#define ioc3_w_midr_r(v) do { ioc3->midr_r = cpu_to_be32(v); } while (0)
-#define ioc3_r_midr_w() be32_to_cpu(ioc3->midr_w)
-#define ioc3_w_midr_w(v) do { ioc3->midr_w = cpu_to_be32(v); } while (0)
-
static inline u32 mcr_pack(u32 pulse, u32 sample)
{
return (pulse << 10) | (sample << 2);
}

-static int nic_wait(struct ioc3 *ioc3)
+static int nic_wait(u32 __iomem *mcr)
{
- u32 mcr;
+ u32 m;

- do {
- mcr = ioc3_r_mcr();
- } while (!(mcr & 2));
+ do {
+ m = readl(mcr);
+ } while (!(m & 2));

- return mcr & 1;
+ return m & 1;
}

-static int nic_reset(struct ioc3 *ioc3)
+static int nic_reset(u32 __iomem *mcr)
{
int presence;

- ioc3_w_mcr(mcr_pack(500, 65));
- presence = nic_wait(ioc3);
+ writel(mcr_pack(500, 65), mcr);
+ presence = nic_wait(mcr);

- ioc3_w_mcr(mcr_pack(0, 500));
- nic_wait(ioc3);
+ writel(mcr_pack(0, 500), mcr);
+ nic_wait(mcr);

return presence;
}

-static inline int nic_read_bit(struct ioc3 *ioc3)
+static inline int nic_read_bit(u32 __iomem *mcr)
{
int result;

- ioc3_w_mcr(mcr_pack(6, 13));
- result = nic_wait(ioc3);
- ioc3_w_mcr(mcr_pack(0, 100));
- nic_wait(ioc3);
+ writel(mcr_pack(6, 13), mcr);
+ result = nic_wait(mcr);
+ writel(mcr_pack(0, 100), mcr);
+ nic_wait(mcr);

return result;
}

-static inline void nic_write_bit(struct ioc3 *ioc3, int bit)
+static inline void nic_write_bit(u32 __iomem *mcr, int bit)
{
if (bit)
- ioc3_w_mcr(mcr_pack(6, 110));
+ writel(mcr_pack(6, 110), mcr);
else
- ioc3_w_mcr(mcr_pack(80, 30));
+ writel(mcr_pack(80, 30), mcr);

- nic_wait(ioc3);
+ nic_wait(mcr);
}

/*
* Read a byte from an iButton device
*/
-static u32 nic_read_byte(struct ioc3 *ioc3)
+static u32 nic_read_byte(u32 __iomem *mcr)
{
u32 result = 0;
int i;

for (i = 0; i < 8; i++)
- result = (result >> 1) | (nic_read_bit(ioc3) << 7);
+ result = (result >> 1) | (nic_read_bit(mcr) << 7);

return result;
}
@@ -285,7 +226,7 @@ static u32 nic_read_byte(struct ioc3 *ioc3)
/*
* Write a byte to an iButton device
*/
-static void nic_write_byte(struct ioc3 *ioc3, int byte)
+static void nic_write_byte(u32 __iomem *mcr, int byte)
{
int i, bit;

@@ -293,23 +234,23 @@ static void nic_write_byte(struct ioc3 *ioc3, int byte)
bit = byte & 1;
byte >>= 1;

- nic_write_bit(ioc3, bit);
+ nic_write_bit(mcr, bit);
}
}

-static u64 nic_find(struct ioc3 *ioc3, int *last)
+static u64 nic_find(u32 __iomem *mcr, int *last)
{
int a, b, index, disc;
u64 address = 0;

- nic_reset(ioc3);
+ nic_reset(mcr);
/* Search ROM. */
- nic_write_byte(ioc3, 0xf0);
+ nic_write_byte(mcr, 0xf0);

/* Algorithm from ``Book of iButton Standards''. */
for (index = 0, disc = 0; index < 64; index++) {
- a = nic_read_bit(ioc3);
- b = nic_read_bit(ioc3);
+ a = nic_read_bit(mcr);
+ b = nic_read_bit(mcr);

if (a && b) {
printk("NIC search failed (not fatal).\n");
@@ -325,14 +266,14 @@ static u64 nic_find(struct ioc3 *ioc3, int *last)
disc = index;
} else if ((address & (1UL << index)) == 0)
disc = index;
- nic_write_bit(ioc3, address & (1UL << index));
+ nic_write_bit(mcr, address & (1UL << index));
continue;
} else {
if (a)
address |= 1UL << index;
else
address &= ~(1UL << index);
- nic_write_bit(ioc3, a);
+ nic_write_bit(mcr, a);
continue;
}
}
@@ -342,7 +283,7 @@ static u64 nic_find(struct ioc3 *ioc3, int *last)
return address;
}

-static int nic_init(struct ioc3 *ioc3)
+static int nic_init(u32 __iomem *mcr)
{
const char *unknown = "unknown";
const char *type = unknown;
@@ -352,7 +293,7 @@ static int nic_init(struct ioc3 *ioc3)

while (1) {
u64 reg;
- reg = nic_find(ioc3, &save);
+ reg = nic_find(mcr, &save);

switch (reg & 0xff) {
case 0x91:
@@ -366,12 +307,12 @@ static int nic_init(struct ioc3 *ioc3)
continue;
}

- nic_reset(ioc3);
+ nic_reset(mcr);

/* Match ROM. */
- nic_write_byte(ioc3, 0x55);
+ nic_write_byte(mcr, 0x55);
for (i = 0; i < 8; i++)
- nic_write_byte(ioc3, (reg >> (i << 3)) & 0xff);
+ nic_write_byte(mcr, (reg >> (i << 3)) & 0xff);

reg >>= 8; /* Shift out type. */
for (i = 0; i < 6; i++) {
@@ -396,15 +337,15 @@ static int nic_init(struct ioc3 *ioc3)
*/
static void ioc3_get_eaddr_nic(struct ioc3_private *ip)
{
- struct ioc3 *ioc3 = ip->regs;
- u8 nic[14];
+ u32 __iomem *mcr = &ip->all_regs->mcr;
int tries = 2; /* There may be some problem with the battery? */
+ u8 nic[14];
int i;

- ioc3_w_gpcr_s(1 << 21);
+ writel(1 << 21, &ip->all_regs->gpcr_s);

while (tries--) {
- if (!nic_init(ioc3))
+ if (!nic_init(mcr))
break;
udelay(500);
}
@@ -415,12 +356,12 @@ static void ioc3_get_eaddr_nic(struct ioc3_private *ip)
}

/* Read Memory. */
- nic_write_byte(ioc3, 0xf0);
- nic_write_byte(ioc3, 0x00);
- nic_write_byte(ioc3, 0x00);
+ nic_write_byte(mcr, 0xf0);
+ nic_write_byte(mcr, 0x00);
+ nic_write_byte(mcr, 0x00);

for (i = 13; i >= 0; i--)
- nic[i] = nic_read_byte(ioc3);
+ nic[i] = nic_read_byte(mcr);

for (i = 2; i < 8; i++)
ip->dev->dev_addr[i - 2] = nic[i];
@@ -441,11 +382,15 @@ static void ioc3_get_eaddr(struct ioc3_private *ip)
static void __ioc3_set_mac_address(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;

- ioc3_w_emar_h((dev->dev_addr[5] << 8) | dev->dev_addr[4]);
- ioc3_w_emar_l((dev->dev_addr[3] << 24) | (dev->dev_addr[2] << 16) |
- (dev->dev_addr[1] << 8) | dev->dev_addr[0]);
+ writel((dev->dev_addr[5] << 8) |
+ dev->dev_addr[4],
+ &ip->regs->emar_h);
+ writel((dev->dev_addr[3] << 24) |
+ (dev->dev_addr[2] << 16) |
+ (dev->dev_addr[1] << 8) |
+ dev->dev_addr[0],
+ &ip->regs->emar_l);
}

static int ioc3_set_mac_address(struct net_device *dev, void *addr)
@@ -469,24 +414,29 @@ static int ioc3_set_mac_address(struct net_device *dev, void *addr)
static int ioc3_mdio_read(struct net_device *dev, int phy, int reg)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;

- while (ioc3_r_micr() & MICR_BUSY);
- ioc3_w_micr((phy << MICR_PHYADDR_SHIFT) | reg | MICR_READTRIG);
- while (ioc3_r_micr() & MICR_BUSY);
+ while (readl(&regs->micr) & MICR_BUSY)
+ ;
+ writel((phy << MICR_PHYADDR_SHIFT) | reg | MICR_READTRIG,
+ &regs->micr);
+ while (readl(&regs->micr) & MICR_BUSY)
+ ;

- return ioc3_r_midr_r() & MIDR_DATA_MASK;
+ return readl(&regs->midr_r) & MIDR_DATA_MASK;
}

static void ioc3_mdio_write(struct net_device *dev, int phy, int reg, int data)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
-
- while (ioc3_r_micr() & MICR_BUSY);
- ioc3_w_midr_w(data);
- ioc3_w_micr((phy << MICR_PHYADDR_SHIFT) | reg);
- while (ioc3_r_micr() & MICR_BUSY);
+ struct ioc3_ethregs *regs = ip->regs;
+
+ while (readl(&regs->micr) & MICR_BUSY)
+ ;
+ writel(data, &regs->midr_w);
+ writel((phy << MICR_PHYADDR_SHIFT) | reg, &regs->micr);
+ while (readl(&regs->micr) & MICR_BUSY)
+ ;
}

static int ioc3_mii_init(struct ioc3_private *ip);
@@ -494,9 +444,9 @@ static int ioc3_mii_init(struct ioc3_private *ip);
static struct net_device_stats *ioc3_get_stats(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;

- dev->stats.collisions += (ioc3_r_etcdc() & ETCDC_COLLCNT_MASK);
+ dev->stats.collisions += readl(&regs->etcdc) & ETCDC_COLLCNT_MASK;
return &dev->stats;
}

@@ -572,7 +522,6 @@ static inline void ioc3_rx(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
struct sk_buff *skb, *new_skb;
- struct ioc3 *ioc3 = ip->regs;
int rx_entry, n_entry, len;
struct ioc3_erxbuf *rxb;
unsigned long *rxr;
@@ -640,7 +589,7 @@ static inline void ioc3_rx(struct net_device *dev)
rxb = (struct ioc3_erxbuf *) (skb->data - RX_OFFSET);
w0 = be32_to_cpu(rxb->w0);
}
- ioc3_w_erpir((n_entry << 3) | ERPIR_ARM);
+ writel((n_entry << 3) | ERPIR_ARM, &ip->regs->erpir);
ip->rx_pi = n_entry;
ip->rx_ci = rx_entry;
}
@@ -648,14 +597,14 @@ static inline void ioc3_rx(struct net_device *dev)
static inline void ioc3_tx(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
+ struct ioc3_ethregs *regs = ip->regs;
unsigned long packets, bytes;
- struct ioc3 *ioc3 = ip->regs;
int tx_entry, o_entry;
struct sk_buff *skb;
u32 etcir;

spin_lock(&ip->ioc3_lock);
- etcir = ioc3_r_etcir();
+ etcir = readl(&regs->etcir);

tx_entry = (etcir >> 7) & 127;
o_entry = ip->tx_ci;
@@ -671,7 +620,7 @@ static inline void ioc3_tx(struct net_device *dev)

o_entry = (o_entry + 1) & 127; /* Next */

- etcir = ioc3_r_etcir(); /* More pkts sent? */
+ etcir = readl(&regs->etcir); /* More pkts sent? */
tx_entry = (etcir >> 7) & 127;
}

@@ -724,44 +673,39 @@ static void ioc3_error(struct net_device *dev, u32 eisr)

/* The interrupt handler does all of the Rx thread work and cleans up
after the Tx thread. */
-static irqreturn_t ioc3_interrupt(int irq, void *_dev)
+static irqreturn_t ioc3_interrupt(int irq, void *dev_id)
{
- struct net_device *dev = (struct net_device *)_dev;
- struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
- const u32 enabled = EISR_RXTIMERINT | EISR_RXOFLO | EISR_RXBUFOFLO |
- EISR_RXMEMERR | EISR_RXPARERR | EISR_TXBUFUFLO |
- EISR_TXEXPLICIT | EISR_TXMEMERR;
+ struct ioc3_private *ip = netdev_priv(dev_id);
+ struct ioc3_ethregs *regs = ip->regs;
u32 eisr;

- eisr = ioc3_r_eisr() & enabled;
-
- ioc3_w_eisr(eisr);
- (void) ioc3_r_eisr(); /* Flush */
+ eisr = readl(&regs->eisr);
+ writel(eisr, &regs->eisr);
+ readl(&regs->eisr); /* Flush */

if (eisr & (EISR_RXOFLO | EISR_RXBUFOFLO | EISR_RXMEMERR |
EISR_RXPARERR | EISR_TXBUFUFLO | EISR_TXMEMERR))
- ioc3_error(dev, eisr);
+ ioc3_error(dev_id, eisr);
if (eisr & EISR_RXTIMERINT)
- ioc3_rx(dev);
+ ioc3_rx(dev_id);
if (eisr & EISR_TXEXPLICIT)
- ioc3_tx(dev);
+ ioc3_tx(dev_id);

return IRQ_HANDLED;
}

static inline void ioc3_setup_duplex(struct ioc3_private *ip)
{
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;

if (ip->mii.full_duplex) {
- ioc3_w_etcsr(ETCSR_FD);
+ writel(ETCSR_FD, &regs->etcsr);
ip->emcr |= EMCR_DUPLEX;
} else {
- ioc3_w_etcsr(ETCSR_HD);
+ writel(ETCSR_HD, &regs->etcsr);
ip->emcr &= ~EMCR_DUPLEX;
}
- ioc3_w_emcr(ip->emcr);
+ writel(ip->emcr, &regs->emcr);
}

static void ioc3_timer(struct timer_list *t)
@@ -936,7 +880,7 @@ static void ioc3_alloc_rings(struct net_device *dev)
static void ioc3_init_rings(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;
unsigned long ring;

ioc3_free_rings(ip);
@@ -947,90 +891,94 @@ static void ioc3_init_rings(struct net_device *dev)

/* Now the rx ring base, consume & produce registers. */
ring = ioc3_map(ip->rxr, 0);
- ioc3_w_erbr_h(ring >> 32);
- ioc3_w_erbr_l(ring & 0xffffffff);
- ioc3_w_ercir(ip->rx_ci << 3);
- ioc3_w_erpir((ip->rx_pi << 3) | ERPIR_ARM);
+ writel(ring >> 32, &regs->erbr_h);
+ writel(ring & 0xffffffff, &regs->erbr_l);
+ writel(ip->rx_ci << 3, &regs->ercir);
+ writel((ip->rx_pi << 3) | ERPIR_ARM, &regs->erpir);

ring = ioc3_map(ip->txr, 0);

ip->txqlen = 0; /* nothing queued */

/* Now the tx ring base, consume & produce registers. */
- ioc3_w_etbr_h(ring >> 32);
- ioc3_w_etbr_l(ring & 0xffffffff);
- ioc3_w_etpir(ip->tx_pi << 7);
- ioc3_w_etcir(ip->tx_ci << 7);
- (void) ioc3_r_etcir(); /* Flush */
+ writel(ring >> 32, &regs->etbr_h);
+ writel(ring & 0xffffffff, &regs->etbr_l);
+ writel(ip->tx_pi << 7, &regs->etpir);
+ writel(ip->tx_ci << 7, &regs->etcir);
+ readl(&regs->etcir); /* Flush */
}

static inline void ioc3_ssram_disc(struct ioc3_private *ip)
{
- struct ioc3 *ioc3 = ip->regs;
- volatile u32 *ssram0 = &ioc3->ssram[0x0000];
- volatile u32 *ssram1 = &ioc3->ssram[0x4000];
- unsigned int pattern = 0x5555;
+ struct ioc3_ethregs *regs = ip->regs;
+ u32 *ssram0 = &ip->ssram[0x0000];
+ u32 *ssram1 = &ip->ssram[0x4000];
+ u32 pattern = 0x5555;

/* Assume the larger size SSRAM and enable parity checking */
- ioc3_w_emcr(ioc3_r_emcr() | (EMCR_BUFSIZ | EMCR_RAMPAR));
+ writel(readl(&regs->emcr) | (EMCR_BUFSIZ | EMCR_RAMPAR), &regs->emcr);
+ readl(&regs->emcr); /* Flush */

- *ssram0 = pattern;
- *ssram1 = ~pattern & IOC3_SSRAM_DM;
+ writel(pattern, ssram0);
+ writel(~pattern & IOC3_SSRAM_DM, ssram1);

- if ((*ssram0 & IOC3_SSRAM_DM) != pattern ||
- (*ssram1 & IOC3_SSRAM_DM) != (~pattern & IOC3_SSRAM_DM)) {
+ if ((readl(ssram0) & IOC3_SSRAM_DM) != pattern ||
+ (readl(ssram1) & IOC3_SSRAM_DM) != (~pattern & IOC3_SSRAM_DM)) {
/* set ssram size to 64 KB */
- ip->emcr = EMCR_RAMPAR;
- ioc3_w_emcr(ioc3_r_emcr() & ~EMCR_BUFSIZ);
- } else
- ip->emcr = EMCR_BUFSIZ | EMCR_RAMPAR;
+ ip->emcr |= EMCR_RAMPAR;
+ writel(readl(&regs->emcr) & ~EMCR_BUFSIZ, &regs->emcr);
+ } else {
+ ip->emcr |= EMCR_BUFSIZ | EMCR_RAMPAR;
+ }
}

static void ioc3_init(struct net_device *dev)
{
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;

del_timer_sync(&ip->ioc3_timer); /* Kill if running */

- ioc3_w_emcr(EMCR_RST); /* Reset */
- (void) ioc3_r_emcr(); /* Flush WB */
+ writel(EMCR_RST, &regs->emcr); /* Reset */
+ readl(&regs->emcr); /* Flush WB */
udelay(4); /* Give it time ... */
- ioc3_w_emcr(0);
- (void) ioc3_r_emcr();
+ writel(0, &regs->emcr);
+ readl(&regs->emcr);

/* Misc registers */
#ifdef CONFIG_SGI_IP27
- ioc3_w_erbar(PCI64_ATTR_BAR >> 32); /* Barrier on last store */
+ /* Barrier on last store */
+ writel(PCI64_ATTR_BAR >> 32, &regs->erbar);
#else
- ioc3_w_erbar(0); /* Let PCI API get it right */
+ /* Let PCI API get it right */
+ writel(0, &regs->erbar);
#endif
- (void) ioc3_r_etcdc(); /* Clear on read */
- ioc3_w_ercsr(15); /* RX low watermark */
- ioc3_w_ertr(0); /* Interrupt immediately */
+ readl(&regs->etcdc); /* Clear on read */
+ writel(15, &regs->ercsr); /* RX low watermark */
+ writel(0, &regs->ertr); /* Interrupt immediately */
__ioc3_set_mac_address(dev);
- ioc3_w_ehar_h(ip->ehar_h);
- ioc3_w_ehar_l(ip->ehar_l);
- ioc3_w_ersr(42); /* XXX should be random */
+ writel(ip->ehar_h, &regs->ehar_h);
+ writel(ip->ehar_l, &regs->ehar_l);
+ writel(42, &regs->ersr); /* XXX should be random */

ioc3_init_rings(dev);

ip->emcr |= ((RX_OFFSET / 2) << EMCR_RXOFF_SHIFT) | EMCR_TXDMAEN |
EMCR_TXEN | EMCR_RXDMAEN | EMCR_RXEN | EMCR_PADEN;
- ioc3_w_emcr(ip->emcr);
- ioc3_w_eier(EISR_RXTIMERINT | EISR_RXOFLO | EISR_RXBUFOFLO |
- EISR_RXMEMERR | EISR_RXPARERR | EISR_TXBUFUFLO |
- EISR_TXEXPLICIT | EISR_TXMEMERR);
- (void) ioc3_r_eier();
+ writel(ip->emcr, &regs->emcr);
+ writel(EISR_RXTIMERINT | EISR_RXOFLO | EISR_RXBUFOFLO |
+ EISR_RXMEMERR | EISR_RXPARERR | EISR_TXBUFUFLO |
+ EISR_TXEXPLICIT | EISR_TXMEMERR, &regs->eier);
+ readl(&regs->eier);
}

static inline void ioc3_stop(struct ioc3_private *ip)
{
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;

- ioc3_w_emcr(0); /* Shutup */
- ioc3_w_eier(0); /* Disable interrupts */
- (void) ioc3_r_eier(); /* Flush */
+ writel(0, &regs->emcr); /* Shutup */
+ writel(0, &regs->eier); /* Disable interrupts */
+ readl(&regs->eier); /* Flush */
}

static int ioc3_open(struct net_device *dev)
@@ -1153,16 +1101,18 @@ static void ioc3_8250_register(struct ioc3_uartregs __iomem *uart)
};
unsigned char lcr;

- lcr = uart->iu_lcr;
- uart->iu_lcr = lcr | UART_LCR_DLAB;
- uart->iu_scr = COSMISC_CONSTANT,
- uart->iu_lcr = lcr;
- uart->iu_lcr;
+ lcr = readb(&uart->iu_lcr);
+ writeb(lcr | UART_LCR_DLAB, &uart->iu_lcr);
+ writeb(COSMISC_CONSTANT, &uart->iu_scr);
+ writeb(lcr, &uart->iu_lcr);
+ readb(&uart->iu_lcr);
serial8250_register_8250_port(&port);
}

static void ioc3_serial_probe(struct pci_dev *pdev, struct ioc3 *ioc3)
{
+ u32 sio_iec;
+
/*
* We need to recognice and treat the fourth MENET serial as it
* does not have an SuperIO chip attached to it, therefore attempting
@@ -1179,29 +1129,31 @@ static void ioc3_serial_probe(struct pci_dev *pdev, struct ioc3 *ioc3)
* Switch IOC3 to PIO mode. It probably already was but let's be
* paranoid
*/
- ioc3->gpcr_s = GPCR_UARTA_MODESEL | GPCR_UARTB_MODESEL;
- ioc3->gpcr_s;
- ioc3->gppr_6 = 0;
- ioc3->gppr_6;
- ioc3->gppr_7 = 0;
- ioc3->gppr_7;
- ioc3->sscr_a = ioc3->sscr_a & ~SSCR_DMA_EN;
- ioc3->sscr_a;
- ioc3->sscr_b = ioc3->sscr_b & ~SSCR_DMA_EN;
- ioc3->sscr_b;
+ writel(GPCR_UARTA_MODESEL | GPCR_UARTB_MODESEL, &ioc3->gpcr_s);
+ readl(&ioc3->gpcr_s);
+ writel(0, &ioc3->gppr[6]);
+ readl(&ioc3->gppr[6]);
+ writel(0, &ioc3->gppr[7]);
+ readl(&ioc3->gppr[7]);
+ writel(readl(&ioc3->port_a.sscr) & ~SSCR_DMA_EN, &ioc3->port_a.sscr);
+ readl(&ioc3->port_a.sscr);
+ writel(readl(&ioc3->port_b.sscr) & ~SSCR_DMA_EN, &ioc3->port_b.sscr);
+ readl(&ioc3->port_b.sscr);
/* Disable all SA/B interrupts except for SA/B_INT in SIO_IEC. */
- ioc3->sio_iec &= ~ (SIO_IR_SA_TX_MT | SIO_IR_SA_RX_FULL |
- SIO_IR_SA_RX_HIGH | SIO_IR_SA_RX_TIMER |
- SIO_IR_SA_DELTA_DCD | SIO_IR_SA_DELTA_CTS |
- SIO_IR_SA_TX_EXPLICIT | SIO_IR_SA_MEMERR);
- ioc3->sio_iec |= SIO_IR_SA_INT;
- ioc3->sscr_a = 0;
- ioc3->sio_iec &= ~ (SIO_IR_SB_TX_MT | SIO_IR_SB_RX_FULL |
- SIO_IR_SB_RX_HIGH | SIO_IR_SB_RX_TIMER |
- SIO_IR_SB_DELTA_DCD | SIO_IR_SB_DELTA_CTS |
- SIO_IR_SB_TX_EXPLICIT | SIO_IR_SB_MEMERR);
- ioc3->sio_iec |= SIO_IR_SB_INT;
- ioc3->sscr_b = 0;
+ sio_iec = readl(&ioc3->sio_iec);
+ sio_iec &= ~(SIO_IR_SA_TX_MT | SIO_IR_SA_RX_FULL |
+ SIO_IR_SA_RX_HIGH | SIO_IR_SA_RX_TIMER |
+ SIO_IR_SA_DELTA_DCD | SIO_IR_SA_DELTA_CTS |
+ SIO_IR_SA_TX_EXPLICIT | SIO_IR_SA_MEMERR);
+ sio_iec |= SIO_IR_SA_INT;
+ sio_iec &= ~(SIO_IR_SB_TX_MT | SIO_IR_SB_RX_FULL |
+ SIO_IR_SB_RX_HIGH | SIO_IR_SB_RX_TIMER |
+ SIO_IR_SB_DELTA_DCD | SIO_IR_SB_DELTA_CTS |
+ SIO_IR_SB_TX_EXPLICIT | SIO_IR_SB_MEMERR);
+ sio_iec |= SIO_IR_SB_INT;
+ writel(sio_iec, &ioc3->sio_iec);
+ writel(0, &ioc3->port_a.sscr);
+ writel(0, &ioc3->port_b.sscr);

ioc3_8250_register(&ioc3->sregs.uarta);
ioc3_8250_register(&ioc3->sregs.uartb);
@@ -1282,7 +1234,9 @@ static int ioc3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
err = -ENOMEM;
goto out_res;
}
- ip->regs = ioc3;
+ ip->regs = &ioc3->eth;
+ ip->ssram = ioc3->ssram;
+ ip->all_regs = ioc3;

#ifdef CONFIG_SERIAL_8250
ioc3_serial_probe(pdev, ioc3);
@@ -1363,12 +1317,11 @@ static void ioc3_remove_one(struct pci_dev *pdev)
{
struct net_device *dev = pci_get_drvdata(pdev);
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;

unregister_netdev(dev);
del_timer_sync(&ip->ioc3_timer);

- iounmap(ioc3);
+ iounmap(ip->all_regs);
pci_release_regions(pdev);
free_netdev(dev);
/*
@@ -1392,11 +1345,10 @@ static struct pci_driver ioc3_driver = {

static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
{
- unsigned long data;
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
- unsigned int len;
struct ioc3_etxd *desc;
+ unsigned long data;
+ unsigned int len;
uint32_t w0 = 0;
int produce;

@@ -1489,7 +1441,7 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
ip->tx_skbs[produce] = skb; /* Remember skb */
produce = (produce + 1) & 127;
ip->tx_pi = produce;
- ioc3_w_etpir(produce << 7); /* Fire ... */
+ writel(produce << 7, &ip->regs->etpir); /* Fire ... */

ip->txqlen++;

@@ -1623,21 +1575,21 @@ static int ioc3_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)

static void ioc3_set_multicast_list(struct net_device *dev)
{
- struct netdev_hw_addr *ha;
struct ioc3_private *ip = netdev_priv(dev);
- struct ioc3 *ioc3 = ip->regs;
+ struct ioc3_ethregs *regs = ip->regs;
+ struct netdev_hw_addr *ha;
u64 ehar = 0;

netif_stop_queue(dev); /* Lock out others. */

if (dev->flags & IFF_PROMISC) { /* Set promiscuous. */
ip->emcr |= EMCR_PROMISC;
- ioc3_w_emcr(ip->emcr);
- (void) ioc3_r_emcr();
+ writel(ip->emcr, &regs->emcr);
+ readl(&regs->emcr);
} else {
ip->emcr &= ~EMCR_PROMISC;
- ioc3_w_emcr(ip->emcr); /* Clear promiscuous. */
- (void) ioc3_r_emcr();
+ writel(ip->emcr, &regs->emcr); /* Clear promiscuous. */
+ readl(&regs->emcr);

if ((dev->flags & IFF_ALLMULTI) ||
(netdev_mc_count(dev) > 64)) {
@@ -1653,8 +1605,8 @@ static void ioc3_set_multicast_list(struct net_device *dev)
ip->ehar_h = ehar >> 32;
ip->ehar_l = ehar & 0xffffffff;
}
- ioc3_w_ehar_h(ip->ehar_h);
- ioc3_w_ehar_l(ip->ehar_l);
+ writel(ip->ehar_h, &regs->ehar_h);
+ writel(ip->ehar_l, &regs->ehar_l);
}

netif_wake_queue(dev); /* Let us get going again. */
--
2.13.7

2019-08-19 16:36:20

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 11/17] net: sgi: ioc3-eth: no need to stop queue set_multicast_list

netif_stop_queue()/netif_wake_qeue() aren't needed for changing
multicast filters. Use spinlocks instead for proper protection
of private struct.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
drivers/net/ethernet/sgi/ioc3-eth.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
index d862f28887f9..7f85a3bfef14 100644
--- a/drivers/net/ethernet/sgi/ioc3-eth.c
+++ b/drivers/net/ethernet/sgi/ioc3-eth.c
@@ -1542,8 +1542,7 @@ static void ioc3_set_multicast_list(struct net_device *dev)
struct netdev_hw_addr *ha;
u64 ehar = 0;

- netif_stop_queue(dev); /* Lock out others. */
-
+ spin_lock_irq(&ip->ioc3_lock);
if (dev->flags & IFF_PROMISC) { /* Set promiscuous. */
ip->emcr |= EMCR_PROMISC;
writel(ip->emcr, &regs->emcr);
@@ -1572,7 +1571,7 @@ static void ioc3_set_multicast_list(struct net_device *dev)
writel(ip->ehar_l, &regs->ehar_l);
}

- netif_wake_queue(dev); /* Let us get going again. */
+ spin_unlock_irq(&ip->ioc3_lock);
}

module_pci_driver(ioc3_driver);
--
2.13.7

2019-08-19 16:36:45

by Thomas Bogendoerfer

[permalink] [raw]
Subject: [PATCH v5 03/17] nvmem: core: add nvmem_device_find

nvmem_device_find provides a way to search for nvmem devices with
the help of a match function simlair to bus_find_device.

Signed-off-by: Thomas Bogendoerfer <[email protected]>
---
Documentation/driver-api/nvmem.rst | 2 ++
drivers/nvmem/core.c | 62 ++++++++++++++++++++------------------
include/linux/nvmem-consumer.h | 9 ++++++
3 files changed, 43 insertions(+), 30 deletions(-)

diff --git a/Documentation/driver-api/nvmem.rst b/Documentation/driver-api/nvmem.rst
index d9d958d5c824..287e86819640 100644
--- a/Documentation/driver-api/nvmem.rst
+++ b/Documentation/driver-api/nvmem.rst
@@ -129,6 +129,8 @@ To facilitate such consumers NVMEM framework provides below apis::
struct nvmem_device *nvmem_device_get(struct device *dev, const char *name);
struct nvmem_device *devm_nvmem_device_get(struct device *dev,
const char *name);
+ struct nvmem_device *nvmem_device_find(void *data,
+ int (*match)(struct device *dev, const void *data));
void nvmem_device_put(struct nvmem_device *nvmem);
int nvmem_device_read(struct nvmem_device *nvmem, unsigned int offset,
size_t bytes, void *buf);
diff --git a/drivers/nvmem/core.c b/drivers/nvmem/core.c
index ac5d945be88a..e591ba54758f 100644
--- a/drivers/nvmem/core.c
+++ b/drivers/nvmem/core.c
@@ -76,36 +76,18 @@ static struct bus_type nvmem_bus_type = {
.name = "nvmem",
};

+#if IS_ENABLED(CONFIG_OF)
static int of_nvmem_match(struct device *dev, const void *nvmem_np)
{
return dev->of_node == nvmem_np;
}
+#endif

-static struct nvmem_device *of_nvmem_find(struct device_node *nvmem_np)
+static int nvmem_match_name(struct device *dev, const void *data)
{
- struct device *d;
-
- if (!nvmem_np)
- return NULL;
-
- d = bus_find_device(&nvmem_bus_type, NULL, nvmem_np, of_nvmem_match);
-
- if (!d)
- return NULL;
+ const char *name = data;

- return to_nvmem_device(d);
-}
-
-static struct nvmem_device *nvmem_find(const char *name)
-{
- struct device *d;
-
- d = bus_find_device_by_name(&nvmem_bus_type, NULL, name);
-
- if (!d)
- return NULL;
-
- return to_nvmem_device(d);
+ return sysfs_streq(name, dev_name(dev));
}

static void nvmem_cell_drop(struct nvmem_cell *cell)
@@ -537,13 +519,16 @@ int devm_nvmem_unregister(struct device *dev, struct nvmem_device *nvmem)
}
EXPORT_SYMBOL(devm_nvmem_unregister);

-static struct nvmem_device *__nvmem_device_get(struct device_node *np,
- const char *nvmem_name)
+static struct nvmem_device *__nvmem_device_get(void *data,
+ int (*match)(struct device *dev, const void *data))
{
struct nvmem_device *nvmem = NULL;
+ struct device *dev;

mutex_lock(&nvmem_mutex);
- nvmem = np ? of_nvmem_find(np) : nvmem_find(nvmem_name);
+ dev = bus_find_device(&nvmem_bus_type, NULL, data, match);
+ if (dev)
+ nvmem = to_nvmem_device(dev);
mutex_unlock(&nvmem_mutex);
if (!nvmem)
return ERR_PTR(-EPROBE_DEFER);
@@ -592,7 +577,7 @@ struct nvmem_device *of_nvmem_device_get(struct device_node *np, const char *id)
if (!nvmem_np)
return ERR_PTR(-ENOENT);

- return __nvmem_device_get(nvmem_np, NULL);
+ return __nvmem_device_get(nvmem_np, of_nvmem_match);
}
EXPORT_SYMBOL_GPL(of_nvmem_device_get);
#endif
@@ -618,10 +603,26 @@ struct nvmem_device *nvmem_device_get(struct device *dev, const char *dev_name)

}

- return __nvmem_device_get(NULL, dev_name);
+ return __nvmem_device_get((void *)dev_name, nvmem_match_name);
}
EXPORT_SYMBOL_GPL(nvmem_device_get);

+/**
+ * nvmem_device_find() - Find nvmem device with matching function
+ *
+ * @data: Data to pass to match function
+ * @match: Callback function to check device
+ *
+ * Return: ERR_PTR() on error or a valid pointer to a struct nvmem_device
+ * on success.
+ */
+struct nvmem_device *nvmem_device_find(void *data,
+ int (*match)(struct device *dev, const void *data))
+{
+ return __nvmem_device_get(data, match);
+}
+EXPORT_SYMBOL_GPL(nvmem_device_find);
+
static int devm_nvmem_device_match(struct device *dev, void *res, void *data)
{
struct nvmem_device **nvmem = res;
@@ -715,7 +716,8 @@ nvmem_cell_get_from_lookup(struct device *dev, const char *con_id)
if ((strcmp(lookup->dev_id, dev_id) == 0) &&
(strcmp(lookup->con_id, con_id) == 0)) {
/* This is the right entry. */
- nvmem = __nvmem_device_get(NULL, lookup->nvmem_name);
+ nvmem = __nvmem_device_get((void *)lookup->nvmem_name,
+ nvmem_match_name);
if (IS_ERR(nvmem)) {
/* Provider may not be registered yet. */
cell = ERR_CAST(nvmem);
@@ -785,7 +787,7 @@ struct nvmem_cell *of_nvmem_cell_get(struct device_node *np, const char *id)
if (!nvmem_np)
return ERR_PTR(-EINVAL);

- nvmem = __nvmem_device_get(nvmem_np, NULL);
+ nvmem = __nvmem_device_get(nvmem_np, of_nvmem_match);
of_node_put(nvmem_np);
if (IS_ERR(nvmem))
return ERR_CAST(nvmem);
diff --git a/include/linux/nvmem-consumer.h b/include/linux/nvmem-consumer.h
index 8f8be5b00060..02dc4aa992b2 100644
--- a/include/linux/nvmem-consumer.h
+++ b/include/linux/nvmem-consumer.h
@@ -89,6 +89,9 @@ void nvmem_del_cell_lookups(struct nvmem_cell_lookup *entries,
int nvmem_register_notifier(struct notifier_block *nb);
int nvmem_unregister_notifier(struct notifier_block *nb);

+struct nvmem_device *nvmem_device_find(void *data,
+ int (*match)(struct device *dev, const void *data));
+
#else

static inline struct nvmem_cell *nvmem_cell_get(struct device *dev,
@@ -204,6 +207,12 @@ static inline int nvmem_unregister_notifier(struct notifier_block *nb)
return -EOPNOTSUPP;
}

+static inline struct nvmem_device *nvmem_device_find(void *data,
+ int (*match)(struct device *dev, const void *data))
+{
+ return NULL;
+}
+
#endif /* CONFIG_NVMEM */

#if IS_ENABLED(CONFIG_NVMEM) && IS_ENABLED(CONFIG_OF)
--
2.13.7

2019-08-19 23:53:19

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v5 00/17] Use MFD framework for SGI IOC3 drivers

On Mon, 19 Aug 2019 18:31:23 +0200, Thomas Bogendoerfer wrote:
> - requested by Jakub I've splitted ioc3 ethernet driver changes into
> more steps to make the transition more visible;

Thanks a lot for doing that!

2019-08-19 23:54:46

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v5 09/17] net: sgi: ioc3-eth: use defines for constants dealing with desc rings

On Mon, 19 Aug 2019 18:31:32 +0200, Thomas Bogendoerfer wrote:
> Descriptor ring sizes of the IOC3 are more or less fixed size. To
> make clearer where there is a relation to ring sizes use defines.
>
> Signed-off-by: Thomas Bogendoerfer <[email protected]>

Reviewed-by: Jakub Kicinski <[email protected]>

2019-08-19 23:56:43

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v5 10/17] net: sgi: ioc3-eth: rework skb rx handling

On Mon, 19 Aug 2019 18:31:33 +0200, Thomas Bogendoerfer wrote:
> Buffers alloacted by alloc_skb() are already cache aligned so there
> is no need for an extra align done by ioc3_alloc_skb. And instead
> of skb_put/skb_trim simply use one skb_put after frame size is known
> during receive.
>
> Signed-off-by: Thomas Bogendoerfer <[email protected]>
> ---
> drivers/net/ethernet/sgi/ioc3-eth.c | 50 ++++++++-----------------------------
> 1 file changed, 11 insertions(+), 39 deletions(-)
>
> diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
> index c875640926d6..d862f28887f9 100644
> --- a/drivers/net/ethernet/sgi/ioc3-eth.c
> +++ b/drivers/net/ethernet/sgi/ioc3-eth.c
> @@ -11,7 +11,6 @@
> *
> * To do:
> *
> - * o Handle allocation failures in ioc3_alloc_skb() more gracefully.
> * o Handle allocation failures in ioc3_init_rings().
> * o Use prefetching for large packets. What is a good lower limit for
> * prefetching?
> @@ -72,6 +71,12 @@
> #define TX_RING_ENTRIES 128
> #define TX_RING_MASK (TX_RING_ENTRIES - 1)
>
> +/* BEWARE: The IOC3 documentation documents the size of rx buffers as
> + * 1644 while it's actually 1664. This one was nasty to track down...
> + */
> +#define RX_OFFSET 10
> +#define RX_BUF_SIZE 1664
> +
> #define ETCSR_FD ((17 << ETCSR_IPGR2_SHIFT) | (11 << ETCSR_IPGR1_SHIFT) | 21)
> #define ETCSR_HD ((21 << ETCSR_IPGR2_SHIFT) | (21 << ETCSR_IPGR1_SHIFT) | 21)
>
> @@ -111,31 +116,6 @@ static void ioc3_init(struct net_device *dev);
> static const char ioc3_str[] = "IOC3 Ethernet";
> static const struct ethtool_ops ioc3_ethtool_ops;
>
> -/* We use this to acquire receive skb's that we can DMA directly into. */
> -
> -#define IOC3_CACHELINE 128UL

Is the cache line on the platform this driver works on 128B?
This looks like a DMA engine alignment requirement, more than an
optimization.

The comment in __alloc_skb() says:

/* We do our best to align skb_shared_info on a separate cache
* line. It usually works because kmalloc(X > SMP_CACHE_BYTES) gives
* aligned memory blocks, unless SLUB/SLAB debug is enabled.
* Both skb->head and skb_shared_info are cache line aligned.
*/

note the "unless".

> -static inline unsigned long aligned_rx_skb_addr(unsigned long addr)
> -{
> - return (~addr + 1) & (IOC3_CACHELINE - 1UL);
> -}
> -
> -static inline struct sk_buff *ioc3_alloc_skb(unsigned long length,
> - unsigned int gfp_mask)
> -{
> - struct sk_buff *skb;
> -
> - skb = alloc_skb(length + IOC3_CACHELINE - 1, gfp_mask);
> - if (likely(skb)) {
> - int offset = aligned_rx_skb_addr((unsigned long)skb->data);
> -
> - if (offset)
> - skb_reserve(skb, offset);
> - }
> -
> - return skb;
> -}
> -
> static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
> {
> #ifdef CONFIG_SGI_IP27
> @@ -148,12 +128,6 @@ static inline unsigned long ioc3_map(void *ptr, unsigned long vdev)
> #endif
> }
>
> -/* BEWARE: The IOC3 documentation documents the size of rx buffers as
> - * 1644 while it's actually 1664. This one was nasty to track down ...
> - */
> -#define RX_OFFSET 10
> -#define RX_BUF_ALLOC_SIZE (1664 + RX_OFFSET + IOC3_CACHELINE)
> -
> #define IOC3_SIZE 0x100000
>
> static inline u32 mcr_pack(u32 pulse, u32 sample)
> @@ -534,10 +508,10 @@ static inline void ioc3_rx(struct net_device *dev)
> err = be32_to_cpu(rxb->err); /* It's valid ... */
> if (err & ERXBUF_GOODPKT) {
> len = ((w0 >> ERXBUF_BYTECNT_SHIFT) & 0x7ff) - 4;
> - skb_trim(skb, len);
> + skb_put(skb, len);
> skb->protocol = eth_type_trans(skb, dev);
>
> - new_skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
> + new_skb = alloc_skb(RX_BUF_SIZE, GFP_ATOMIC);
> if (!new_skb) {
> /* Ouch, drop packet and just recycle packet
> * to keep the ring filled.
> @@ -546,6 +520,7 @@ static inline void ioc3_rx(struct net_device *dev)
> new_skb = skb;
> goto next;
> }
> + new_skb->dev = dev;

Assigning dev pointer seems unrelated to the rest of the patch?

> if (likely(dev->features & NETIF_F_RXCSUM))
> ioc3_tcpudp_checksum(skb,
> @@ -556,8 +531,6 @@ static inline void ioc3_rx(struct net_device *dev)
>
> ip->rx_skbs[rx_entry] = NULL; /* Poison */
>
> - /* Because we reserve afterwards. */
> - skb_put(new_skb, (1664 + RX_OFFSET));
> rxb = (struct ioc3_erxbuf *)new_skb->data;
> skb_reserve(new_skb, RX_OFFSET);
>
> @@ -846,16 +819,15 @@ static void ioc3_alloc_rings(struct net_device *dev)
> for (i = 0; i < RX_BUFFS; i++) {
> struct sk_buff *skb;
>
> - skb = ioc3_alloc_skb(RX_BUF_ALLOC_SIZE, GFP_ATOMIC);
> + skb = alloc_skb(RX_BUF_SIZE, GFP_ATOMIC);
> if (!skb) {
> show_free_areas(0, NULL);
> continue;
> }
> + skb->dev = dev;
>
> ip->rx_skbs[i] = skb;
>
> - /* Because we reserve afterwards. */
> - skb_put(skb, (1664 + RX_OFFSET));
> rxb = (struct ioc3_erxbuf *)skb->data;
> rxr[i] = cpu_to_be64(ioc3_map(rxb, 1));
> skb_reserve(skb, RX_OFFSET);

2019-08-20 00:06:13

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v5 11/17] net: sgi: ioc3-eth: no need to stop queue set_multicast_list

On Mon, 19 Aug 2019 18:31:34 +0200, Thomas Bogendoerfer wrote:
> netif_stop_queue()/netif_wake_qeue() aren't needed for changing
> multicast filters. Use spinlocks instead for proper protection
> of private struct.
>
> Signed-off-by: Thomas Bogendoerfer <[email protected]>
> ---
> drivers/net/ethernet/sgi/ioc3-eth.c | 5 ++---
> 1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/ethernet/sgi/ioc3-eth.c b/drivers/net/ethernet/sgi/ioc3-eth.c
> index d862f28887f9..7f85a3bfef14 100644
> --- a/drivers/net/ethernet/sgi/ioc3-eth.c
> +++ b/drivers/net/ethernet/sgi/ioc3-eth.c
> @@ -1542,8 +1542,7 @@ static void ioc3_set_multicast_list(struct net_device *dev)
> struct netdev_hw_addr *ha;
> u64 ehar = 0;
>
> - netif_stop_queue(dev); /* Lock out others. */
> -
> + spin_lock_irq(&ip->ioc3_lock);

What does this lock protect? ???? No question that stopping TX queues
makes little sense, but this function is only called from
ndo_set_rx_mode(), so with rtnl_lock held.

I thought it may protect ip->emcr, but that one is accessed with no
locking from the ioc3_timer() -> ioc3_setup_duplex() path..

> if (dev->flags & IFF_PROMISC) { /* Set promiscuous. */
> ip->emcr |= EMCR_PROMISC;
> writel(ip->emcr, &regs->emcr);
> @@ -1572,7 +1571,7 @@ static void ioc3_set_multicast_list(struct net_device *dev)
> writel(ip->ehar_l, &regs->ehar_l);
> }
>
> - netif_wake_queue(dev); /* Let us get going again. */
> + spin_unlock_irq(&ip->ioc3_lock);
> }
>
> module_pci_driver(ioc3_driver);

2019-08-20 00:08:53

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v5 12/17] net: sgi: ioc3-eth: use dma-direct for dma allocations

On Mon, 19 Aug 2019 18:31:35 +0200, Thomas Bogendoerfer wrote:
> @@ -1386,18 +1427,24 @@ static netdev_tx_t ioc3_start_xmit(struct sk_buff *skb, struct net_device *dev)
> unsigned long b2 = (data | 0x3fffUL) + 1UL;
> unsigned long s1 = b2 - data;
> unsigned long s2 = data + len - b2;
> + dma_addr_t d;
>
> desc->cmd = cpu_to_be32(len | ETXD_INTWHENDONE |
> ETXD_B1V | ETXD_B2V | w0);
> desc->bufcnt = cpu_to_be32((s1 << ETXD_B1CNT_SHIFT) |
> (s2 << ETXD_B2CNT_SHIFT));
> - desc->p1 = cpu_to_be64(ioc3_map(skb->data, 1));
> - desc->p2 = cpu_to_be64(ioc3_map((void *)b2, 1));
> + d = dma_map_single(ip->dma_dev, skb->data, s1, DMA_TO_DEVICE);

You'll need to check the DMA address with dma_mapping_error(dev, addr),
otherwise static checkers will get upset.

> + desc->p1 = cpu_to_be64(ioc3_map(d, PCI64_ATTR_PREF));
> + d = dma_map_single(ip->dma_dev, (void *)b2, s1, DMA_TO_DEVICE);
> + desc->p2 = cpu_to_be64(ioc3_map(d, PCI64_ATTR_PREF));

2019-08-21 14:56:55

by Thomas Bogendoerfer

[permalink] [raw]
Subject: Re: [PATCH v5 11/17] net: sgi: ioc3-eth: no need to stop queue set_multicast_list

On Mon, 19 Aug 2019 17:04:53 -0700
Jakub Kicinski <[email protected]> wrote:

> On Mon, 19 Aug 2019 18:31:34 +0200, Thomas Bogendoerfer wrote:
> > netif_stop_queue()/netif_wake_qeue() aren't needed for changing
> > multicast filters. Use spinlocks instead for proper protection
> > of private struct.
> >
>
> I thought it may protect ip->emcr, but that one is accessed with no
> locking from the ioc3_timer() -> ioc3_setup_duplex() path..

it should protect ip->emcr ... I'll add spin_lock/unlock to setup_duplex and
respin the patch.

Thomas.

--
SUSE Linux GmbH
GF: Felix Imend?rffer, Mary Higgins, Sri Rasiah
HRB 21284 (AG N?rnberg)

2019-08-21 15:38:42

by Thomas Bogendoerfer

[permalink] [raw]
Subject: Re: [PATCH v5 10/17] net: sgi: ioc3-eth: rework skb rx handling

On Mon, 19 Aug 2019 16:55:22 -0700
Jakub Kicinski <[email protected]> wrote:

> On Mon, 19 Aug 2019 18:31:33 +0200, Thomas Bogendoerfer wrote:
> > Buffers alloacted by alloc_skb() are already cache aligned so there
> > is no need for an extra align done by ioc3_alloc_skb. And instead
> > of skb_put/skb_trim simply use one skb_put after frame size is known
> > during receive.
> [...]
> > -/* We use this to acquire receive skb's that we can DMA directly into. */
> > -
> > -#define IOC3_CACHELINE 128UL
>
> Is the cache line on the platform this driver works on 128B?

right now yes, in theory IOC3 CAD DUO cards might work in SGI O2 systems,
but the current Linux PCI implementation for O2 will not detect that card.
On X86 usually the BIOS will choke up on that cards.

> This looks like a DMA engine alignment requirement, more than an
> optimization.

that true, there are two constraints for the rx buffers, start must be aligned
to 128 bytes and a buffer must not cross a 16kbyte boundary. I was already
thinking of allocating pages and chop them up. Is there a Linux API available,
which could help for implementing this ?

I'll probably drop this patch or only change the skb_put stuff plus RX_BUF_SIZE
define.

Thomas.

--
SUSE Linux GmbH
GF: Felix Imend?rffer, Mary Higgins, Sri Rasiah
HRB 21284 (AG N?rnberg)

2019-08-21 20:39:11

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v5 10/17] net: sgi: ioc3-eth: rework skb rx handling

On Wed, 21 Aug 2019 16:28:47 +0200, Thomas Bogendoerfer wrote:
> > This looks like a DMA engine alignment requirement, more than an
> > optimization.
>
> that true, there are two constraints for the rx buffers, start must be aligned
> to 128 bytes and a buffer must not cross a 16kbyte boundary. I was already
> thinking of allocating pages and chop them up. Is there a Linux API available,
> which could help for implementing this ?
>
> I'll probably drop this patch or only change the skb_put stuff plus RX_BUF_SIZE
> define.

Sounds a little like frag allocator (napi_alloc_frag()/
netdev_alloc_frag()), but I'm not sure you'd have sufficient control
to skip over the 16k boundary.. Perhaps others have better suggestions.

2019-09-03 09:12:03

by Paul Burton

[permalink] [raw]
Subject: Re: [PATCH v5 04/17] MIPS: PCI: refactor ioc3 special handling

Hello,

Thomas Bogendoerfer wrote:
> Refactored code to only have one ioc3 special handling for read
> access and one for write access.

Applied to mips-next.

> commit 813cafc4109c
> https://git.kernel.org/mips/c/813cafc4109c
>
> Signed-off-by: Thomas Bogendoerfer <[email protected]>
> Signed-off-by: Paul Burton <[email protected]>

Thanks,
Paul

[ This message was auto-generated; if you believe anything is incorrect
then please email [email protected] to report it. ]