net: phytium: update phytium net driver support to 6.6.0.4#1679
Open
mamingrui123 wants to merge 13 commits into
Open
net: phytium: update phytium net driver support to 6.6.0.4#1679mamingrui123 wants to merge 13 commits into
mamingrui123 wants to merge 13 commits into
Conversation
The tx_ptr_lock is used in both softirq context (via macb_tx_complete in
NAPI poll) and task context (via macb_mac_link_down in workqueue).
Currently, macb_tx_complete() uses spin_lock(), which is appropriate for
softirq context, while macb_mac_link_down() also uses spin_lock(), which
does **not** disable softirqs. This mismatch causes a potential deadlock
if softirq interrupts the task context while it holds the lock, as
detected by lockdep:
WARNING: inconsistent lock state
inconsistent {IN-SOFTIRQ-W} -> {SOFTIRQ-ON-W} usage.
...
*** DEADLOCK ***
Fix this by replacing spin_lock/unlock with spin_lock_bh/unlock_bh in
macb_mac_link_down(), ensuring softirqs are disabled before acquiring
the lock in task context.
Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn>
Signed-off-by: Li Wencheng <liwencheng@phytium.com.cn>
Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
The driver was using plain memcpy() to copy data to/from __iomem msg_regs. When the MMIO address is not naturally aligned (e.g. ...904a), this triggers an alignment fault. With KASAN enabled, this fault is reliably reproducible during driver probe: FSC = 0x21: alignment fault pc : __memcpy+0x58/0x230 phytmac_v2_get_mac_addr+0xdc/0x180 [phytmac] Root cause: using memcpy() on MMIO regions both violates I/O access semantics and can lead to unaligned accesses. Fix it by replacing memcpy() with memcpy_fromio() when reading messages from msg ring and memcpy_toio() when writing messages to the msg ring. This avoids unaligned wide accesses and ensures proper MMIO semantics. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Li Wencheng <liwencheng@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
On certain machines, the phy_id cannot be read. Adjusting the MDC setting to occur before registering the mdiobus in enable_mdio_control resolves this issue. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Ouyang Chenxing <ouyangchenxing2110@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Change the NULL return value to ERR_PTR(-EAGAIN) and handle it properly in the upper-layer function to break the processing loop, thereby avoiding warnings. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Ouyang Chenxing <ouyangchenxing2110@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Add support for NETIF_F_TSO6 in features check and default config to enable IPv6 TSO functionality. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Ouyang Chenxing <ouyangchenxing2110@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Add NETIF_F_TSO6 support to MACB_NETIF_LSO to enable IPv6 TSO functionality. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Ouyang Chenxing <ouyangchenxing2110@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Add support for the Phytium PCIe endpoint virtual Ethernet (EP VNet) driver. This driver enables the endpoint side of a PCIe-based virtual Ethernet link, typically paired with a corresponding root complex (RC) driver. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Add support for the Phytium PCIe root complex virtual Ethernet (RC VNet) driver. This driver enables the root complex side of a PCIe-based virtual Ethernet link, typically paired with a corresponding endpoint (EP) driver. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
To ensure proper auto-negotiation with the PHY at different link speeds, the MAC requires mode-specific PHY initialization during speed transitions. This patch adds dynamic speed reconfiguration for SGMII mode, allowing the driver to switch between 2.5G/1G/100M/10M as needed. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
isabling CONFIG_ACPI will cause errors like 'error: implicit declaration of function ‘acpi_evaluate_integer’, so it is necessary to add #ifdef CONFIG_ACPI for protection. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Disabling CONFIG_ACPI will cause errors like 'error: implicit declaration of function ‘acpi_get_acpi_dev’', so it is necessary to add #ifdef CONFIG_ACPI for protection. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
…ing bug Fix CRC errors observed on the EP side during high-rate UDP transmission. Two issues in the RC TX path could lead to EP DMA reading stale or incorrect data: - The TX reclaim threshold condition was never met, causing descriptors to accumulate and TX queue state to become inconsistent. - Incorrect TX index management allowed descriptors to be reused before their payloads were safely reclaimed. Fix both the reclaim condition and index handling to ensure TX descriptors are reclaimed in time and payload/DMA address mapping remains correct to preventing EP-side CRC mismatches. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Fix error handling paths in pci_epf_vnet_probe() to avoid unregistering an unregistered char device when alloc_chrdev_region() fails. Also add a NULL check for vnet in pci_epf_vnet_bind() to prevent NULL pointer dereference when bind is called after a failed probe. Signed-off-by: Ma Mingrui <mamingrui1243@phytium.com.cn> Signed-off-by: Li Tongfeng <litongfeng1497@phytium.com.cn> Signed-off-by: Wang Yinfeng <wangyinfeng@phytium.com.cn>
Reviewer's GuideUpdates the Phytium Ethernet MAC driver and adds a new PCIe-based virtual network driver stack, along with a few fixes and capability tweaks across Phytium and Cadence MAC drivers and related LED/ACPI handling. Sequence diagram for packet transmit from RC to EP over PCIe virtual networksequenceDiagram
participant App as RC_Application
participant NetRC as RC_netdev(pci_rc_vnet)
participant TxQRC as RC_tx_queue
participant PCI as PCIe_link
participant EPC as EP_DMA_engine(pci_epc)
participant NetEP as EP_netdev(pci_epf_vnet)
participant RxQEP as EP_rx_queue
App->>NetRC: send_skb()
NetRC->>NetRC: pci_rc_vnet_start_xmit(skb)
NetRC->>TxQRC: allocate_desc_and_set_addr_crc_len
NetRC->>PCI: dma_map_single_and_sync
NetRC->>TxQRC: mark_desc_USED
NetRC->>NetRC: update_tx_tail
NetRC->>PCI: write(SEND_MSI_IRQ_to_BAR1)
PCI-->>NetEP: interrupt_rx_irqhandler
NetEP->>NetEP: pci_epf_vnet_rx_irqhandler()
NetEP->>NetEP: napi_schedule(rx_napi)
NetEP->>NetEP: pci_epf_vnet_rx_poll(budget)
NetEP->>RxQEP: read_head_tail_and_desc
NetEP->>EPC: pci_epf_vnet_rc2ep_dma(dst_page_addr,src_rc_addr,len)
EPC-->>NetEP: DMA_READ_complete_or_timeout
NetEP->>NetEP: crc32_check_and_error_handling
NetEP->>NetEP: ep_build_skb()
NetEP->>RxQEP: clear_desc_USED
NetEP->>NetEP: netif_receive_skb(skb)
NetEP-->>App: packet_delivered_to_stack
Class diagram for new PCIe EP virtual net driver structuresclassDiagram
class pci_epf_vnet {
+u32 msg_enable
+void* reg[MAX_BAR_NUM]
+struct pci_epf* epf
+enum pci_barno vnet_reg_barno
+struct hrtimer timer
+struct timer_list link_detect_timer
+u32 rx_refill_start
+unsigned int rx_refill_num
+u32 rx_ring_size
+u32 rx_buffer_size
+struct sk_buff** rx_skb
+struct ep_rx_buffer* rx_buffer_info
+int rx_buffer_len
+u32 rx_next_to_alloc
+struct param_range rxd_size
+u32 tx_head
+u32 tx_tail
+u32 tx_ring_size
+struct param_range txd_size
+struct ep_tx_buffer* tx_buffer_info
+struct net_device* netdev
+spinlock_t tx_lock
+spinlock_t rx_lock
+struct napi_struct rx_napi
+struct pci_ep_queue* queue
+int(*clean_rc2ep)(struct pci_epf_vnet* vnet,int budget)
+int rx_irq
+bool bind_success
}
class pci_ep_queue {
+struct ep_queue ep_rx_queue
+struct ep_queue ep_tx_queue
}
class ep_queue {
+u32 irq_num
+u32 tail
+u32 head
+u32 nb_desc
+u32 flags
+struct pci_ep_dma_desc dma_desc_base[VNET_MAX_RXD]
}
class pci_ep_dma_desc {
+u64 addr
+u64 ctrl
}
class ep_rx_buffer {
+dma_addr_t addr
+struct page* page
+u16 page_offset
+u16 pagecnt_bias
}
class ep_tx_buffer {
+dma_addr_t addr
+void* vaddr
+struct page* page
+u16 page_offset
+u16 pagecnt_bias
}
class phytium_pcie_ep {
+void __iomem* reg_base
+struct resource* mem_res
+void __iomem* hpb_base
+unsigned int max_regions
+unsigned long ob_region_map
+phys_addr_t* ob_addr
+phys_addr_t irq_phys_addr
+void __iomem* irq_cpu_addr
+unsigned long irq_pci_addr
+u8 irq_pci_fn
+struct pci_epc* epc
+struct platform_device* pdev
}
class cdev_mdata {
+dev_t dev_num
+struct page* page
+phys_addr_t phys_addr
+struct cdev* epf_cdev
+struct class* epf_class
+struct device* epf_device
+struct pci_epf_vnet* cdev_vnet
}
class epf_ioctl_data {
+u64 ktext_addr
+u32 ktext_len
}
pci_epf_vnet --> pci_ep_queue : queue
pci_ep_queue --> ep_queue : ep_rx_queue
pci_ep_queue --> ep_queue : ep_tx_queue
ep_queue --> pci_ep_dma_desc : dma_desc_base
pci_epf_vnet --> ep_rx_buffer : rx_buffer_info
pci_epf_vnet --> ep_tx_buffer : tx_buffer_info
phytium_pcie_ep --> pci_epf_vnet : used_by_epf_vnet
cdev_mdata --> pci_epf_vnet : cdev_vnet
Class diagram for new PCIe RC virtual net driver structuresclassDiagram
class pci_rc_vnet_private {
+u32 msg_enable
+int rx_buffer_len
+struct pci_dev* pci_dev
+void __iomem* bar[BAR_NUM]
+dma_addr_t* tx_phys_addr_list
+struct sk_buff** tx_skbuff
+struct rc_rx_buffer* rx_buffer_info
+u32 rx_next_to_alloc
+struct pci_rc_vnet_queue* queue
+struct pci_rc_vnet_tx_queue* tx_queue
+struct pci_rc_vnet_rx_queue* rx_queue
+struct net_device* netdev
+int irq_type
+int num_irqs
+spinlock_t tx_lock
+spinlock_t tx_reclaim_lock
+spinlock_t rx_lock
+struct params params
+struct work_struct tx_timeout_task
+int(*clean_ep2rc)(struct pci_rc_vnet_private* tp,int budget)
+struct napi_struct tx_napi
+struct napi_struct rx_napi
+int tx_reclaim_start
+int tx_reclaim_num
+int rx_refill_start
+struct hrtimer link_detect_timer
+struct hrtimer receive_timer
+struct hrtimer tx_poll_timer
+void __iomem* msi_irq_addr
}
class pci_rc_vnet_queue {
+struct pci_rc_vnet_tx_queue tx_queue
+struct pci_rc_vnet_rx_queue rx_queue
}
class pci_rc_vnet_tx_queue {
+u32 irq
+u32 tail
+u32 head
+u32 nb_desc
+u32 flags
+struct pci_rc_vnet_dma_desc desc_ring[PCI_RC_VNET_DMA_BUFF_NUM]
}
class pci_rc_vnet_rx_queue {
+u32 irq
+u32 tail
+u32 head
+u32 nb_desc
+u32 flags
+struct pci_rc_vnet_dma_desc desc_ring[PCI_RC_VNET_DMA_BUFF_NUM]
}
class pci_rc_vnet_dma_desc {
+u64 addr
+u64 ctrl
}
class rc_rx_buffer {
+dma_addr_t addr
+struct page* page
+u16 page_offset
+u16 pagecnt_bias
}
class params {
+struct param_range rx_ring
+struct param_range tx_ring
}
class param_range {
+u32 min
+u32 max
+u32 count
}
pci_rc_vnet_private --> pci_rc_vnet_queue : queue
pci_rc_vnet_queue --> pci_rc_vnet_tx_queue : tx_queue
pci_rc_vnet_queue --> pci_rc_vnet_rx_queue : rx_queue
pci_rc_vnet_tx_queue --> pci_rc_vnet_dma_desc : desc_ring
pci_rc_vnet_rx_queue --> pci_rc_vnet_dma_desc : desc_ring
pci_rc_vnet_private --> rc_rx_buffer : rx_buffer_info
pci_rc_vnet_private --> params : params
Class diagram for updated Phytium MAC driver structuresclassDiagram
class phyinit_cfg {
+u64 phy_domain
+u64 phy_multiplex
+u64 phy_mode
+u64 phy_speed
}
class phytmac_hw_if {
+int(*reset_hw)(struct phytmac* pdata)
+int(*powerup_hw)(struct phytmac* pdata,int on)
+int(*mac_linkup)(struct phytmac* pdata,phy_interface_t interface,int speed,int duplex)
+void(*enable_pause)(struct phytmac* pdata,int rx_pause)
+void(*disable_tx)(struct phytmac* pdata)
+void(*disable_rx)(struct phytmac* pdata)
+void(*phy_speed_switch)(struct phytmac* pdata,int speed)
+int(*enable_mdio_control)(struct phytmac* pdata,int enable)
+int(*mdio_read)(struct phytmac* pdata,int mii_id,int regnum)
+int(*mdio_write)(struct phytmac* pdata,int mii_id,int regnum,u16 value)
}
class phytmac {
+struct net_device* ndev
+struct device* dev
+struct phytmac_hw_if* hw_if
+int pause
+phy_interface_t phy_interface
+enum phytmac_interface phytmac_v2_interface
+int ori_speed
+int speed
+int duplex
+int autoneg
+unsigned long capacities
+spinlock_t lock
}
class phytmac_1p0_hw {
+struct phytmac_hw_if phytmac_1p0_hw
}
phytmac_hw_if <.. phytmac_1p0_hw : instance
phytmac --> phytmac_hw_if : hw_if
phytmac_1p0_hw ..> phyinit_cfg : uses_for_smc
File-Level Changes
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
Hi @mamingrui123. Thanks for your PR. I'm waiting for a deepin-community member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hey - I've found 1 issue, and left some high level feedback:
- The new multi-buffer RX handling in phytmac_rx() should use IS_ERR()/PTR_ERR() when interpreting the return from phytmac_rx_mbuffer(), otherwise applying PTR_ERR() unconditionally to a valid skb pointer is unsafe and could misbehave if phytmac_rx_mbuffer() ever returns other ERR_PTR codes.
- The EP virtual net driver relies on a single global epf_cdev_mdata (including the cdev, page and phys_addr) for its char-device interface; consider making this state per-EPF instance to avoid conflicts or incorrect behaviour when multiple endpoints are present.
- phytmac_phy_speed_switch() matches MAC instances via hard-coded physical base addresses (MAC0_ADDR_BASE/MAC1_ADDR_BASE) taken from ndev->base_addr; this is fragile if the mapping changes or the driver is reused on other SoCs, so it would be more robust to key the MAC ID off driver-internal configuration rather than fixed physical constants.
Prompt for AI Agents
Please address the comments from this code review:
## Overall Comments
- The new multi-buffer RX handling in phytmac_rx() should use IS_ERR()/PTR_ERR() when interpreting the return from phytmac_rx_mbuffer(), otherwise applying PTR_ERR() unconditionally to a valid skb pointer is unsafe and could misbehave if phytmac_rx_mbuffer() ever returns other ERR_PTR codes.
- The EP virtual net driver relies on a single global epf_cdev_mdata (including the cdev, page and phys_addr) for its char-device interface; consider making this state per-EPF instance to avoid conflicts or incorrect behaviour when multiple endpoints are present.
- phytmac_phy_speed_switch() matches MAC instances via hard-coded physical base addresses (MAC0_ADDR_BASE/MAC1_ADDR_BASE) taken from ndev->base_addr; this is fragile if the mapping changes or the driver is reused on other SoCs, so it would be more robust to key the MAC ID off driver-internal configuration rather than fixed physical constants.
## Individual Comments
### Comment 1
<location path="drivers/net/ethernet/phytium-pci-vnet/pci-ep-net/pci_epf_vnet.c" line_range="1198" />
<code_context>
+ u64 new_len = size + offset;
+ u64 read_len = 0;
+ u8 *ori_addr = bmcaddr;
+ void *vaddr = memremap(epf_cdev_mdata.phys_addr, MEM_MAX, MEMREMAP_WB);
+
+ while (new_len) {
</code_context>
<issue_to_address>
**issue (bug_risk):** memremap is being used on a DMA address, not a CPU physical address, which is not generally valid.
`epf_cdev_mdata.phys_addr` comes from `dma_map_page_attrs`, so it’s a DMA address, not a guaranteed CPU physical address (especially with an IOMMU). Using `memremap` on it is unsafe. To get a CPU-mappable address for `memremap`, derive a true physical address from the page (e.g. `page_to_phys(epf_cdev_mdata.page)`), or better, avoid `memremap` and reuse the existing kernel virtual mapping of the allocated page instead.
</issue_to_address>Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.
| u64 new_len = size + offset; | ||
| u64 read_len = 0; | ||
| u8 *ori_addr = bmcaddr; | ||
| void *vaddr = memremap(epf_cdev_mdata.phys_addr, MEM_MAX, MEMREMAP_WB); |
There was a problem hiding this comment.
issue (bug_risk): memremap is being used on a DMA address, not a CPU physical address, which is not generally valid.
epf_cdev_mdata.phys_addr comes from dma_map_page_attrs, so it’s a DMA address, not a guaranteed CPU physical address (especially with an IOMMU). Using memremap on it is unsafe. To get a CPU-mappable address for memremap, derive a true physical address from the page (e.g. page_to_phys(epf_cdev_mdata.page)), or better, avoid memremap and reuse the existing kernel virtual mapping of the allocated page instead.
There was a problem hiding this comment.
Pull request overview
This PR updates Phytium Ethernet support and adds a new Phytium PCIe virtual networking stack (EP/RC) intended for PCIe-based datapath configurations, while also refining feature advertisement, link handling, and some platform/ACPI guards.
Changes:
- Extend Phytium MAC support (SGMII speed switching hook, RX multi-buffer handling, feature updates including TSO6, and MMIO-safe message ring access).
- Introduce new Phytium PCI virtual Ethernet drivers for Root Complex and Endpoint sides, plus Kconfig/Makefile integration.
- Minor related network driver updates (MACB TSO6 advertisement + link-down TX cleanup) and ACPI-guarded LED initialization.
Reviewed changes
Copilot reviewed 19 out of 19 changed files in this pull request and generated 14 comments.
Show a summary per file
| File | Description |
|---|---|
| drivers/net/ethernet/phytium/phytmac.h | Bump driver version; add SMC/PHY init definitions; add ori_speed and hw hook for PHY speed switching. |
| drivers/net/ethernet/phytium/phytmac_v2.c | Use memcpy_{to,from}io for MMIO rings; adjust MDC handling; CONFIG_ACPI guards; SGMII autoneg toggle fix. |
| drivers/net/ethernet/phytium/phytmac_v1.h | Add MAC base address constants used for SMCCC speed switching selection. |
| drivers/net/ethernet/phytium/phytmac_v1.c | Add SMCCC-based PHY speed switching and plumb it into hw interface; CONFIG_ACPI guard; SGMII autoneg toggle fix. |
| drivers/net/ethernet/phytium/phytmac_main.c | Fix RX multi-buffer loop behavior with -EAGAIN; add SGMII speed-switch plumbing; extend TSO alignment checks and advertise TSO6. |
| drivers/net/ethernet/phytium-pci-vnet/Kconfig | Add menu and vendor option for Phytium PCI virtual Ethernet drivers. |
| drivers/net/ethernet/phytium-pci-vnet/Makefile | Build integration for EP and RC subdrivers. |
| drivers/net/ethernet/phytium-pci-vnet/pci-rc-net/Kconfig | Add RC-side virtual Ethernet Kconfig option. |
| drivers/net/ethernet/phytium-pci-vnet/pci-rc-net/Makefile | Build integration for RC driver object. |
| drivers/net/ethernet/phytium-pci-vnet/pci-rc-net/pci_rc_vnet.h | Add RC driver interface/structures/constants. |
| drivers/net/ethernet/phytium-pci-vnet/pci-rc-net/pci_rc_vnet.c | Add RC-side virtual netdev implementation (DMA queues + timers + ethtool). |
| drivers/net/ethernet/phytium-pci-vnet/pci-ep-net/Kconfig | Add EP-side virtual Ethernet Kconfig option. |
| drivers/net/ethernet/phytium-pci-vnet/pci-ep-net/Makefile | Build integration for EP driver object. |
| drivers/net/ethernet/phytium-pci-vnet/pci-ep-net/pci_epf_vnet.h | Add EPF driver interface/structures/constants, including cdev interface. |
| drivers/net/ethernet/phytium-pci-vnet/pci-ep-net/pci_epf_vnet.c | Add EP-side virtual netdev + EPF binding + DMA engine interactions + cdev/ioctl path. |
| drivers/net/ethernet/Makefile | Wire Phytium vendor directory and new Phytium PCI vnet directory into ethernet build. |
| drivers/net/ethernet/Kconfig | Source new Phytium and Phytium PCI vnet Kconfigs (but currently duplicates xircom include). |
| drivers/net/ethernet/cadence/macb_main.c | Advertise TSO6 alongside TSO; use spin_lock_bh in link-down TX cleanup. |
| drivers/char/phytnetled/phytnet_led.c | Guard ACPI-only LED/netdev init paths with CONFIG_ACPI. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| if (pdata->xdp_prog) | ||
| netdev_warn(pdata->ndev, "xdp does not support multiple buffers!!\n"); | ||
| skb = phytmac_rx_mbuffer(queue); | ||
| if (PTR_ERR(skb) == -EAGAIN) { |
Comment on lines
168
to
173
| source "drivers/net/ethernet/pensando/Kconfig" | ||
| source "drivers/net/ethernet/phytium/Kconfig" | ||
| source "drivers/net/ethernet/phytium-pci-vnet/Kconfig" | ||
| source "drivers/net/ethernet/qlogic/Kconfig" | ||
| source "drivers/net/ethernet/xircom/Kconfig" | ||
| source "drivers/net/ethernet/brocade/Kconfig" |
| hrtimer_start(&tp->receive_timer, ns_to_ktime(PCI_RC_VNET_RECEIVE_PERIOD_NS), | ||
| HRTIMER_MODE_REL_SOFT); | ||
| hrtimer_start(&tp->tx_poll_timer, ns_to_ktime(PCI_RC_VNET_TX_POLL_PERIOD_NS), | ||
| HRTIMER_MODE_REL_SOFT); |
Comment on lines
+543
to
+581
| static int pci_rc_vnet_rx(struct pci_rc_vnet_private *tp, u32 head, u32 tail) | ||
| { | ||
| u32 desc_crc, pkt_crc; | ||
| int used; | ||
| u64 ctrl; | ||
| int pkt_len; | ||
| void *pkt_data; | ||
|
|
||
| struct rc_rx_buffer *rx_buffer_info; | ||
| struct pci_rc_vnet_rx_queue *rx_queue; | ||
| struct pci_rc_vnet_dma_desc *rx_desc_ring; | ||
| struct sk_buff *skb; | ||
| int ring_size = tp->params.tx_ring.count; | ||
| unsigned int pkt_num, count = 0; | ||
|
|
||
| rx_queue = tp->rx_queue; | ||
| rx_desc_ring = rx_queue->desc_ring; | ||
|
|
||
| if (tail >= head) | ||
| pkt_num = tail - head; | ||
| else | ||
| pkt_num = ring_size - head + tail; | ||
|
|
||
| for (count = 0; count < pkt_num; count++) { | ||
| /* Ensure ctrl is at least as up-to-date as rxused */ | ||
| dma_rmb(); | ||
|
|
||
| ctrl = rx_desc_ring[head].ctrl; | ||
| used = PCI_RC_VNET_DMA_DESC_FIELD_GET(USED, ctrl); | ||
|
|
||
| pkt_len = PCI_RC_VNET_DMA_DESC_FIELD_GET(PKT_LEN, ctrl); | ||
|
|
||
| /* No package to receive */ | ||
| if (!used) | ||
| break; | ||
|
|
||
| /* get rx buffer */ | ||
| rx_buffer_info = &tp->rx_buffer_info[head & (ring_size - 1)]; | ||
| if (!rx_buffer_info->page) { |
Comment on lines
+663
to
+671
| static void pci_rc_vnet_get_strings(struct net_device *netdev, u32 stringset, u8 *data) | ||
| { | ||
| switch (stringset) { | ||
| case ETH_SS_TEST: | ||
| memcpy(data, *pci_rc_vnet_gstrings_test, sizeof(pci_rc_vnet_gstrings_test)); | ||
| break; | ||
| case ETH_SS_STATS: | ||
| memcpy(data, *pci_rc_vnet_gstrings_stats, sizeof(pci_rc_vnet_gstrings_stats)); | ||
| break; |
| EP_TX_PAGE_SIZE, | ||
| DMA_FROM_DEVICE, EP_DMA_ATTR); | ||
| if (dma_mapping_error(epc->dev.parent, paddr)) { | ||
| netdev_err(vnet->netdev, "tx map page failed\n"); |
| int timeout; | ||
| unsigned long lock_flags; | ||
|
|
||
| if (len < 0 || len > MEM_MAX) |
Comment on lines
+1134
to
+1142
| static int epf_cdev_mmap(struct file *filp, struct vm_area_struct *vma) | ||
| { | ||
| unsigned long pfn = epf_cdev_mdata.phys_addr >> PAGE_SHIFT; | ||
| unsigned long size = vma->vm_end - vma->vm_start; | ||
|
|
||
| if (remap_pfn_range(vma, vma->vm_start, pfn, size, vma->vm_page_prot)) | ||
| return -EAGAIN; | ||
|
|
||
| return 0; |
| continue; | ||
|
|
||
| dma_unmap_page_attrs(epc->dev.parent, tx_buffer_info->addr, | ||
| EP_RX_PAGE_SIZE, DMA_FROM_DEVICE, EP_DMA_ATTR); |
Comment on lines
+1196
to
+1202
| u64 read_len = 0; | ||
| u8 *ori_addr = bmcaddr; | ||
| void *vaddr = memremap(epf_cdev_mdata.phys_addr, MEM_MAX, MEMREMAP_WB); | ||
|
|
||
| while (new_len) { | ||
| if (new_len >= MEM_MAX) | ||
| read_len = MEM_MAX; |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This patches updates the support for phytium net driver.
Summary by Sourcery
Update Phytium Ethernet support with SGMII speed switching, improved feature handling, and add a new PCI-based virtual network driver for EP/RC configurations.
New Features:
Bug Fixes:
Enhancements:
Build: