[PATCH 11/14] hw/net: GMAC Rx Implementation

[Nabih Estefan]

From: Nabih Estefan Diaz <nabihestefan@google.com>

- Implementation of Receive function for packets
- Implementation for reading and writing from and to descriptors in
  memory for Rx

NOTE: At this point in development we believe this function is working
as intended, and the kernel supports these findings, but we need the
Transmit function to work before we upload

Signed-off-by: Nabih Estefan Diaz <nabihestefan@google.com>

hw/net: npcm_gmac Flush queued packets when starting RX

When RX starts, we need to flush the queued packets so that they
can be received by the GMAC device. Without this it won't work
with TAP NIC device.

Signed-off-by: Hao Wu <wuhaotsh@google.com>

hw/net: Handle RX desc full in NPCM GMAC

When RX descriptor list is full, it returns a DMA_STATUS for software to handle 
it. But there's no way to indicate the software ha handled all RX descriptors 
and the whole pipeline stalls.

We do something similar to NPCM7XX EMC to handle this case.

1. Return packet size when RX descriptor is full, effectively dropping these 
packets in such a case.
2. When software clears RX descriptor full bit, continue receiving further 
packets by flushing QEMU packet queue.

Signed-off-by: Hao Wu <wuhaotsh@google.com>

hw/net: Receive and drop packets when descriptors are full in GMAC

Effectively this allows QEMU to receive and drop incoming packets when
RX descriptors are full. Similar to EMC, this lets GMAC to drop packets
faster, especially during bootup sequence.

Signed-off-by: Hao Wu <wuhaotsh@google.com>
---
 hw/net/npcm_gmac.c | 353 +++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 324 insertions(+), 29 deletions(-)

diff --git a/hw/net/npcm_gmac.c b/hw/net/npcm_gmac.c
index 6f8109e0ee..67f123e3c4 100644
--- a/hw/net/npcm_gmac.c
+++ b/hw/net/npcm_gmac.c
@@ -23,7 +23,11 @@
 #include "hw/registerfields.h"
 #include "hw/net/mii.h"
 #include "hw/net/npcm_gmac.h"
+#include "linux/if_ether.h"
 #include "migration/vmstate.h"
+#include "net/checksum.h"
+#include "net/net.h"
+#include "qemu/cutils.h"
 #include "qemu/log.h"
 #include "qemu/units.h"
 #include "sysemu/dma.h"
@@ -91,7 +95,6 @@ REG32(NPCM_GMAC_PTP_TTSR, 0x71c)
 #define NPCM_DMA_BUS_MODE_SWR               BIT(0)
 
 static const uint32_t npcm_gmac_cold_reset_values[NPCM_GMAC_NR_REGS] = {
-    /* Reduce version to 3.2 so that the kernel can enable interrupt. */
     [R_NPCM_GMAC_VERSION]         = 0x00001032,
     [R_NPCM_GMAC_TIMER_CTRL]      = 0x03e80000,
     [R_NPCM_GMAC_MAC0_ADDR_HI]    = 0x8000ffff,
@@ -146,6 +149,17 @@ static void gmac_phy_set_link(NPCMGMACState *s, bool 
active)
 
 static bool gmac_can_receive(NetClientState *nc)
 {
+    NPCMGMACState *gmac = NPCM_GMAC(qemu_get_nic_opaque(nc));
+
+    /* If GMAC receive is disabled. */
+    if (!(gmac->regs[R_NPCM_GMAC_MAC_CONFIG] & NPCM_GMAC_MAC_CONFIG_RX_EN)) {
+        return false;
+    }
+
+    /* If GMAC DMA RX is stopped. */
+    if (!(gmac->regs[R_NPCM_DMA_CONTROL] & NPCM_DMA_CONTROL_START_STOP_RX)) {
+        return false;
+    }
     return true;
 }
 
@@ -191,11 +205,285 @@ static void gmac_update_irq(NPCMGMACState *gmac)
     qemu_set_irq(gmac->irq, level);
 }
 
-static ssize_t gmac_receive(NetClientState *nc, const uint8_t *buf, size_t len)
+static int gmac_read_rx_desc(dma_addr_t addr, struct NPCMGMACRxDesc *desc)
 {
-    /* Placeholder */
+    if (dma_memory_read(&address_space_memory, addr, desc,
+                        sizeof(*desc), MEMTXATTRS_UNSPECIFIED)) {
+        qemu_log_mask(LOG_GUEST_ERROR, "%s: Failed to read descriptor @ 0x%"
+                      HWADDR_PRIx "\n", __func__, addr);
+        return -1;
+    }
+    desc->rdes0 = le32_to_cpu(desc->rdes0);
+    desc->rdes1 = le32_to_cpu(desc->rdes1);
+    desc->rdes2 = le32_to_cpu(desc->rdes2);
+    desc->rdes3 = le32_to_cpu(desc->rdes3);
+    return 0;
+}
+
+static int gmac_write_rx_desc(dma_addr_t addr, struct NPCMGMACRxDesc *desc)
+{
+    struct NPCMGMACRxDesc le_desc;
+    le_desc.rdes0 = cpu_to_le32(desc->rdes0);
+    le_desc.rdes1 = cpu_to_le32(desc->rdes1);
+    le_desc.rdes2 = cpu_to_le32(desc->rdes2);
+    le_desc.rdes3 = cpu_to_le32(desc->rdes3);
+    if (dma_memory_write(&address_space_memory, addr, &le_desc,
+                        sizeof(le_desc), MEMTXATTRS_UNSPECIFIED)) {
+        qemu_log_mask(LOG_GUEST_ERROR, "%s: Failed to write descriptor @ 0x%"
+                      HWADDR_PRIx "\n", __func__, addr);
+        return -1;
+    }
+    return 0;
+}
+
+static int gmac_read_tx_desc(dma_addr_t addr, struct NPCMGMACTxDesc *desc)
+{
+    if (dma_memory_read(&address_space_memory, addr, desc,
+                        sizeof(*desc), MEMTXATTRS_UNSPECIFIED)) {
+        qemu_log_mask(LOG_GUEST_ERROR, "%s: Failed to read descriptor @ 0x%"
+                      HWADDR_PRIx "\n", __func__, addr);
+        return -1;
+    }
+    desc->tdes0 = le32_to_cpu(desc->tdes0);
+    desc->tdes1 = le32_to_cpu(desc->tdes1);
+    desc->tdes2 = le32_to_cpu(desc->tdes2);
+    desc->tdes3 = le32_to_cpu(desc->tdes3);
+    return 0;
+}
+
+static int gmac_write_tx_desc(dma_addr_t addr, struct NPCMGMACTxDesc *desc)
+{
+    struct NPCMGMACTxDesc le_desc;
+    le_desc.tdes0 = cpu_to_le32(desc->tdes0);
+    le_desc.tdes1 = cpu_to_le32(desc->tdes1);
+    le_desc.tdes2 = cpu_to_le32(desc->tdes2);
+    le_desc.tdes3 = cpu_to_le32(desc->tdes3);
+    if (dma_memory_write(&address_space_memory, addr, &le_desc,
+                        sizeof(le_desc), MEMTXATTRS_UNSPECIFIED)) {
+        qemu_log_mask(LOG_GUEST_ERROR, "%s: Failed to write descriptor @ 0x%"
+                      HWADDR_PRIx "\n", __func__, addr);
+        return -1;
+    }
     return 0;
 }
+static int gmac_rx_transfer_frame_to_buffer(uint32_t rx_buf_len,
+                                            uint32_t *left_frame,
+                                            uint32_t rx_buf_addr,
+                                            bool *eof_transferred,
+                                            const uint8_t *frame_ptr,
+                                            uint16_t *transferred)
+{
+    uint32_t to_transfer;
+    /*
+     * Check that buffer is bigger than the frame being transfered
+     * If bigger then transfer only whats left of frame
+     * Else, fill frame with all the content possible
+     */
+    if (rx_buf_len >= *left_frame) {
+        to_transfer = *left_frame;
+        *eof_transferred = true;
+    } else {
+        to_transfer = rx_buf_len;
+    }
+
+    /* write frame part to memory */
+    if (dma_memory_write(&address_space_memory, (uint64_t) rx_buf_addr,
+                         frame_ptr, to_transfer, MEMTXATTRS_UNSPECIFIED))
+    {
+        return -1;
+    }
+
+    /* update frame pointer and size of whats left of frame */
+    frame_ptr += to_transfer;
+    *left_frame -= to_transfer;
+    *transferred += to_transfer;
+
+    return 0;
+}
+
+static void gmac_dma_set_state(NPCMGMACState *gmac, int shift, uint32_t state)
+{
+    gmac->regs[R_NPCM_DMA_STATUS] = deposit32(gmac->regs[R_NPCM_DMA_STATUS],
+        shift, 3, state);
+}
+
+static ssize_t gmac_receive(NetClientState *nc, const uint8_t *buf, size_t len)
+{
+    /*
+     * Comments have steps that relate to the
+     * receiving process steps in pg 386
+     */
+    NPCMGMACState *gmac = NPCM_GMAC(qemu_get_nic_opaque(nc));
+    uint32_t left_frame = len;
+    const uint8_t *frame_ptr = buf;
+    uint32_t desc_addr;
+    uint32_t rx_buf_len, rx_buf_addr;
+    struct NPCMGMACRxDesc rx_desc;
+    uint16_t transferred = 0;
+    bool eof_transferred = false;
+
+    trace_npcm_gmac_packet_receive(DEVICE(gmac)->canonical_path, len);
+    if (!gmac_can_receive(nc)) {
+        qemu_log_mask(LOG_GUEST_ERROR, "GMAC Currently is not able for Rx");
+        return -1;
+    }
+    if (!gmac->regs[R_NPCM_DMA_HOST_RX_DESC]) {
+        gmac->regs[R_NPCM_DMA_HOST_RX_DESC] =
+            NPCM_DMA_HOST_RX_DESC_MASK(gmac->regs[R_NPCM_DMA_RX_BASE_ADDR]);
+    }
+    desc_addr = 
NPCM_DMA_HOST_RX_DESC_MASK(gmac->regs[R_NPCM_DMA_HOST_RX_DESC]);
+
+    /* step 1 */
+    gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+        NPCM_DMA_STATUS_RX_RUNNING_FETCHING_STATE);
+    trace_npcm_gmac_packet_desc_read(DEVICE(gmac)->canonical_path, desc_addr);
+    if (gmac_read_rx_desc(desc_addr, &rx_desc)) {
+        qemu_log_mask(LOG_GUEST_ERROR, "RX Descriptor @ 0x%x cant be read\n",
+                      desc_addr);
+        gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+            NPCM_DMA_STATUS_RX_SUSPENDED_STATE);
+        return -1;
+    }
+
+    /* step 2 */
+    if (!(rx_desc.rdes0 & RX_DESC_RDES0_OWN)) {
+        qemu_log_mask(LOG_GUEST_ERROR,
+                      "RX Descriptor @ 0x%x is owned by software\n",
+                      desc_addr);
+        gmac->regs[R_NPCM_DMA_STATUS] |= NPCM_DMA_STATUS_RU;
+        gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+            NPCM_DMA_STATUS_RX_SUSPENDED_STATE);
+        gmac_update_irq(gmac);
+        return len;
+    }
+    /* step 3 */
+    /*
+     * TODO --
+     * Implement all frame filtering and processing (with its own interrupts)
+     */
+    trace_npcm_gmac_debug_desc_data(DEVICE(gmac)->canonical_path, &rx_desc,
+                                    rx_desc.rdes0, rx_desc.rdes1, 
rx_desc.rdes2,
+                                    rx_desc.rdes3);
+    /* Set FS in first descriptor */
+    rx_desc.rdes0 |= RX_DESC_RDES0_FIRST_DESC_MASK;
+
+    gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+        NPCM_DMA_STATUS_RX_RUNNING_TRANSFERRING_STATE);
+
+    /* Pad the frame with FCS as the kernel driver will strip it away. */
+    left_frame += ETH_FCS_LEN;
+
+    /* repeat while we still have frame to transfer to memory */
+    while (!eof_transferred) {
+        /* Return descriptor no matter what happens */
+        rx_desc.rdes0 &= ~RX_DESC_RDES0_OWN;
+        /* Set the frame to be an IPv4/IPv6 frame. */
+        rx_desc.rdes0 |= RX_DESC_RDES0_FRM_TYPE_MASK;
+
+        /* step 4 */
+        rx_buf_len = RX_DESC_RDES1_BFFR1_SZ_MASK(rx_desc.rdes1);
+        rx_buf_addr = rx_desc.rdes2;
+        gmac->regs[R_NPCM_DMA_CUR_RX_BUF_ADDR] = rx_buf_addr;
+        gmac_rx_transfer_frame_to_buffer(rx_buf_len, &left_frame, rx_buf_addr,
+                                         &eof_transferred, frame_ptr,
+                                         &transferred);
+
+        trace_npcm_gmac_packet_receiving_buffer(DEVICE(gmac)->canonical_path,
+                                                rx_buf_len, rx_buf_addr);
+        /* if we still have frame left and the second buffer is not chained */
+         if (!(rx_desc.rdes1 & RX_DESC_RDES1_SEC_ADDR_CHND_MASK) && \
+              !eof_transferred) {
+            /* repeat process from above on buffer 2 */
+            rx_buf_len = RX_DESC_RDES1_BFFR2_SZ_MASK(rx_desc.rdes1);
+            rx_buf_addr = rx_desc.rdes3;
+            gmac->regs[R_NPCM_DMA_CUR_RX_BUF_ADDR] = rx_buf_addr;
+            gmac_rx_transfer_frame_to_buffer(rx_buf_len, &left_frame,
+                                             rx_buf_addr, &eof_transferred,
+                                             frame_ptr, &transferred);
+            trace_npcm_gmac_packet_receiving_buffer( \
+                                                DEVICE(gmac)->canonical_path,
+                                                rx_buf_len, rx_buf_addr);
+        }
+        /* update address for descriptor */
+        gmac->regs[R_NPCM_DMA_HOST_RX_DESC] = rx_buf_addr;
+        /* Return descriptor */
+        rx_desc.rdes0 &= ~RX_DESC_RDES0_OWN;
+        /* Update frame length transferred */
+        rx_desc.rdes0 |= ((uint32_t)transferred)
+            << RX_DESC_RDES0_FRAME_LEN_SHIFT;
+        trace_npcm_gmac_debug_desc_data(DEVICE(gmac)->canonical_path, &rx_desc,
+                                        rx_desc.rdes0, rx_desc.rdes1,
+                                        rx_desc.rdes2, rx_desc.rdes3);
+
+        /* step 5 */
+        gmac_write_rx_desc(desc_addr, &rx_desc);
+        trace_npcm_gmac_debug_desc_data(DEVICE(gmac)->canonical_path,
+                                        &rx_desc, rx_desc.rdes0,
+                                        rx_desc.rdes1, rx_desc.rdes2,
+                                        rx_desc.rdes3);
+        /* read new descriptor into rx_desc if needed*/
+        if (!eof_transferred) {
+            /* Get next descriptor address (chained or sequential) */
+            if (rx_desc.rdes1 & RX_DESC_RDES1_RC_END_RING_MASK) {
+                desc_addr = gmac->regs[R_NPCM_DMA_RX_BASE_ADDR];
+            } else if (rx_desc.rdes1 & RX_DESC_RDES1_SEC_ADDR_CHND_MASK) {
+                desc_addr = rx_desc.rdes3;
+            } else {
+                desc_addr += sizeof(rx_desc);
+            }
+            trace_npcm_gmac_packet_desc_read(DEVICE(gmac)->canonical_path,
+                                             desc_addr);
+            if (gmac_read_rx_desc(desc_addr, &rx_desc)) {
+                qemu_log_mask(LOG_GUEST_ERROR,
+                              "RX Descriptor @ 0x%x cant be read\n",
+                              desc_addr);
+                gmac->regs[R_NPCM_DMA_STATUS] |= NPCM_DMA_STATUS_RU;
+                gmac_update_irq(gmac);
+                return len;
+            }
+
+            /* step 6 */
+            if (rx_desc.rdes0 & RX_DESC_RDES0_OWN) {
+                if (!(gmac->regs[R_NPCM_DMA_CONTROL] & \
+                     NPCM_DMA_CONTROL_FLUSH_MASK)) {
+                    rx_desc.rdes0 |= RX_DESC_RDES0_DESC_ERR_MASK;
+                }
+                eof_transferred = true;
+            }
+        }
+    }
+    gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+        NPCM_DMA_STATUS_RX_RUNNING_CLOSING_STATE);
+
+    rx_desc.rdes0 |= RX_DESC_RDES0_LAST_DESC_MASK;
+    if (!(rx_desc.rdes1 & RX_DESC_RDES1_DIS_INTR_COMP_MASK)) {
+        gmac->regs[R_NPCM_DMA_STATUS] |= NPCM_DMA_STATUS_RI;
+        gmac_update_irq(gmac);
+    }
+    trace_npcm_gmac_debug_desc_data(DEVICE(gmac)->canonical_path, &rx_desc,
+                                    rx_desc.rdes0, rx_desc.rdes1, 
rx_desc.rdes2,
+                                    rx_desc.rdes3);
+
+    /* step 8 */
+    gmac->regs[R_NPCM_DMA_CONTROL] |= NPCM_DMA_CONTROL_FLUSH_MASK;
+
+    /* step 9 */
+    trace_npcm_gmac_packet_received(DEVICE(gmac)->canonical_path, left_frame);
+    gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+        NPCM_DMA_STATUS_RX_RUNNING_WAITING_STATE);
+    gmac_write_rx_desc(desc_addr, &rx_desc);
+
+    /* Get next descriptor address (chained or sequential) */
+    if (rx_desc.rdes1 & RX_DESC_RDES1_RC_END_RING_MASK) {
+        desc_addr = gmac->regs[R_NPCM_DMA_RX_BASE_ADDR];
+    } else if (rx_desc.rdes1 & RX_DESC_RDES1_SEC_ADDR_CHND_MASK) {
+        desc_addr = rx_desc.rdes3;
+    } else {
+        desc_addr += sizeof(rx_desc);
+    }
+    gmac->regs[R_NPCM_DMA_HOST_RX_DESC] = desc_addr;
+    return len;
+}
 static void gmac_cleanup(NetClientState *nc)
 {
     /* Nothing to do yet. */
@@ -281,7 +569,6 @@ static void npcm_gmac_write(void *opaque, hwaddr offset,
                               uint64_t v, unsigned size)
 {
     NPCMGMACState *gmac = opaque;
-    uint32_t prev;
 
     trace_npcm_gmac_reg_write(DEVICE(gmac)->canonical_path, offset, v);
 
@@ -305,22 +592,7 @@ static void npcm_gmac_write(void *opaque, hwaddr offset,
         break;
 
     case A_NPCM_GMAC_MAC_CONFIG:
-        prev = gmac->regs[offset / sizeof(uint32_t)];
         gmac->regs[offset / sizeof(uint32_t)] = v;
-
-        /* If transmit is being enabled for first time, update desc addr */
-        if (~(prev & NPCM_GMAC_MAC_CONFIG_TX_EN) &
-             (v & NPCM_GMAC_MAC_CONFIG_TX_EN)) {
-            gmac->regs[R_NPCM_DMA_HOST_TX_DESC] =
-                gmac->regs[R_NPCM_DMA_TX_BASE_ADDR];
-        }
-
-        /* If receive is being enabled for first time, update desc addr */
-        if (~(prev & NPCM_GMAC_MAC_CONFIG_RX_EN) &
-             (v & NPCM_GMAC_MAC_CONFIG_RX_EN)) {
-            gmac->regs[R_NPCM_DMA_HOST_RX_DESC] =
-                gmac->regs[R_NPCM_DMA_RX_BASE_ADDR];
-        }
         break;
 
     case A_NPCM_GMAC_MII_ADDR:
@@ -362,6 +634,31 @@ static void npcm_gmac_write(void *opaque, hwaddr offset,
 
     case A_NPCM_DMA_RCV_POLL_DEMAND:
         /* We dont actually care about the value */
+        gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+            NPCM_DMA_STATUS_RX_RUNNING_WAITING_STATE);
+        break;
+
+    case A_NPCM_DMA_XMT_POLL_DEMAND:
+        /* We dont actually care about the value */
+        gmac_try_send_next_packet(gmac);
+        break;
+
+    case A_NPCM_DMA_CONTROL:
+        gmac->regs[offset / sizeof(uint32_t)] = v;
+        if (v & NPCM_DMA_CONTROL_START_STOP_TX) {
+            gmac_try_send_next_packet(gmac);
+        } else {
+            gmac_dma_set_state(gmac, NPCM_DMA_STATUS_TX_PROCESS_STATE_SHIFT,
+                NPCM_DMA_STATUS_TX_STOPPED_STATE);
+        }
+        if (v & NPCM_DMA_CONTROL_START_STOP_RX) {
+            gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+                NPCM_DMA_STATUS_RX_RUNNING_WAITING_STATE);
+            qemu_flush_queued_packets(qemu_get_queue(gmac->nic));
+        } else {
+            gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+                NPCM_DMA_STATUS_RX_STOPPED_STATE);
+        }
         break;
 
     case A_NPCM_DMA_STATUS:
@@ -371,16 +668,14 @@ static void npcm_gmac_write(void *opaque, hwaddr offset,
                           "%s: Write of read-only bits of reg: offset: 0x%04"
                            HWADDR_PRIx ", value: 0x%04" PRIx64 "\n",
                            DEVICE(gmac)->canonical_path, offset, v);
-        } else {
-            /* for W1c bits, implement W1C */
-            gmac->regs[offset / sizeof(uint32_t)] &=
-                ~NPCM_DMA_STATUS_W1C_MASK(v);
-            if (v & NPCM_DMA_STATUS_NIS_BITS) {
-                gmac->regs[offset / sizeof(uint32_t)] &= ~NPCM_DMA_STATUS_NIS;
-            }
-            if (v & NPCM_DMA_STATUS_AIS_BITS) {
-                gmac->regs[offset / sizeof(uint32_t)] &= ~NPCM_DMA_STATUS_AIS;
-            }
+        }
+        /* for W1C bits, implement W1C */
+        gmac->regs[offset / sizeof(uint32_t)] &= ~NPCM_DMA_STATUS_W1C_MASK(v);
+        if (v & NPCM_DMA_STATUS_RU) {
+            /* Clearing RU bit indicates descriptor is owned by DMA again. */
+            gmac_dma_set_state(gmac, NPCM_DMA_STATUS_RX_PROCESS_STATE_SHIFT,
+                NPCM_DMA_STATUS_RX_RUNNING_WAITING_STATE);
+            qemu_flush_queued_packets(qemu_get_queue(gmac->nic));
         }
         break;
 
-- 
2.42.0.459.ge4e396fd5e-goog