On Mon, Aug 14, 2023 at 07:32:48PM +0600, Maxim Uvarov wrote:
> Implement network lwIP interface connected to the U-boot.
> Keep original file structure widely used fro lwIP ports.
> (i.e. port/if.c port/sys-arch.c).
What the patch does is obvious. Try to describe *why* we need this
>
> Signed-off-by: Maxim Uvarov <maxim.uvarov@linaro.org>
> ---
> net/eth-uclass.c | 8 +
> net/lwip/port/if.c | 260 ++++++++++++++++++++++++++
> net/lwip/port/include/arch/cc.h | 39 ++++
> net/lwip/port/include/arch/sys_arch.h | 56 ++++++
> net/lwip/port/include/limits.h | 0
> net/lwip/port/sys-arch.c | 20 ++
> 6 files changed, 383 insertions(+)
> create mode 100644 net/lwip/port/if.c
> create mode 100644 net/lwip/port/include/arch/cc.h
> create mode 100644 net/lwip/port/include/arch/sys_arch.h
> create mode 100644 net/lwip/port/include/limits.h
> create mode 100644 net/lwip/port/sys-arch.c
>
> diff --git a/net/eth-uclass.c b/net/eth-uclass.c
> index c393600fab..6625f6d8a5 100644
> --- a/net/eth-uclass.c
> +++ b/net/eth-uclass.c
> @@ -32,6 +32,7 @@ DECLARE_GLOBAL_DATA_PTR;
> struct eth_device_priv {
> enum eth_state_t state;
> bool running;
> + ulwip ulwip;
> };
>
> /**
> @@ -347,6 +348,13 @@ int eth_init(void)
> return ret;
> }
>
> +struct ulwip *eth_lwip_priv(struct udevice *current)
> +{
> + struct eth_device_priv *priv = dev_get_uclass_priv(current);
> +
> + return &priv->ulwip;
> +}
> +
> void eth_halt(void)
> {
> struct udevice *current;
> diff --git a/net/lwip/port/if.c b/net/lwip/port/if.c
> new file mode 100644
> index 0000000000..625a9c10bf
> --- /dev/null
> +++ b/net/lwip/port/if.c
> @@ -0,0 +1,260 @@
> +// SPDX-License-Identifier: GPL-2.0
> +
> +/*
> + * (C) Copyright 2023 Linaro Ltd. <maxim.uvarov@linaro.org>
> + */
> +
> +#include <common.h>
> +#include <command.h>
> +#include <net/eth.h>
> +#include "lwip/debug.h"
> +#include "lwip/arch.h"
> +#include "netif/etharp.h"
> +#include "lwip/stats.h"
> +#include "lwip/def.h"
> +#include "lwip/mem.h"
> +#include "lwip/pbuf.h"
> +#include "lwip/sys.h"
> +#include "lwip/netif.h"
> +#include "lwip/ethip6.h"
> +
> +#include "lwip/ip.h"
> +
> +#define IFNAME0 'e'
> +#define IFNAME1 '0'
Why is this needed and how was 'e0' chosen?
Dont we have a device name in the udevice struct?
If we assume that we have lwip netif on top of an active U-Boot eth device, then it can be any name.
/** descriptive abbreviation */
char name[2];
./net/lwip/lwip-external/contrib/examples/ethernetif/ethernetif.c:#define IFNAME0 'e'
./net/lwip/lwip-external/contrib/examples/ethernetif/ethernetif.c:#define IFNAME1 'n'
./net/lwip/lwip-external/contrib/ports/unix/port/netif/tapif.c:#define IFNAME0 't'
./net/lwip/lwip-external/contrib/ports/unix/port/netif/tapif.c:#define IFNAME1 'p'
./net/lwip/lwip-external/contrib/ports/unix/port/netif/vdeif.c:#define IFNAME0 'v'
./net/lwip/lwip-external/contrib/ports/unix/port/netif/vdeif.c:#define IFNAME1 'd'
./net/lwip/lwip-external/contrib/ports/win32/pcapif.c:#define IFNAME0 'e'
./net/lwip/lwip-external/contrib/ports/win32/pcapif.c:#define IFNAME1 '0'
./net/lwip/lwip-external/src/netif/bridgeif.c:#define IFNAME0 'b'
./net/lwip/lwip-external/src/netif/bridgeif.c:#define IFNAME1 'r'
./net/lwip/port/if.c:#define IFNAME0 'u'
./net/lwip/port/if.c:#define IFNAME1 '0'
> +
> +static struct pbuf *low_level_input(struct netif *netif);
> +
> +int ulwip_enabled(void)
> +{
> + struct ulwip *ulwip;
> +
> + ulwip = eth_lwip_priv(eth_get_dev());
eth_get_dev() can return NULL. There are various locations of this call
that needs fixing
ok.
> + return ulwip->init_done;
> +}
> +
> +
> +struct ulwip_if {
> +};
Why the forward declaration?
> +
> +#define LWIP_PORT_INIT_NETMASK(addr) IP4_ADDR((addr), 255, 255, 255, 0)
Why are we limiting the netmask to a class C network?
that can be completely removed.
> +
> +void ulwip_poll(void)
> +{
> + struct pbuf *p;
> + int err;
> + struct netif *netif = netif_get_by_index(1);
First of all netif can be NULL. Apart from that always requesting index 1
feels wrong. We should do something similar to eth_get_dev() and get the
*active* device correlation to an index
I expect that it will be 1 lwip netif for all eth defs. And only active eth dev works with lwip.
This can be reconsidered but might be not a part of the initial patch set.
> +
> + p = low_level_input(netif);
> + if (!p) {
> + log_err("error p = low_level_input = NULL\n");
This looks like a debug message.
'Network interface undefined' or something else, which is more readable.
> + return;
> + }
> +
> + /* ethernet_input always returns ERR_OK */
> + err = ethernet_input(p, netif);
> + if (err)
> + log_err("ip4_input err %d\n", err);
> +
> + return;
> +}
> +
> +static struct pbuf *low_level_input(struct netif *netif)
> +{
> + struct pbuf *p, *q;
> + u16_t len = net_rx_packet_len;
> + uchar *data = "">
> +
> + /* We allocate a pbuf chain of pbufs from the pool. */
> + p = pbuf_alloc(PBUF_RAW, len, PBUF_POOL);
> + if (p) {
if (!p) and reverse the logic
> + /* We iterate over the pbuf chain until we have read the entire
> + * packet into the pbuf.
> + */
> + for (q = p; q != NULL; q = q->next) {
> + /*
> + * Read enough bytes to fill this pbuf in the chain. The
> + * available data in the pbuf is given by the q->len
> + * variable.
> + * This does not necessarily have to be a memcpy, you can also preallocate
> + * pbufs for a DMA-enabled MAC and after receiving truncate it to the
> + * actually received size. In this case, ensure the tot_len member of the
> + * pbuf is the sum of the chained pbuf len members.
> + */
> + MEMCPY(q->payload, data, q->len);
> + data += q->len;
> + }
> + // acknowledge that packet has been read();
> +
> + LINK_STATS_INC(link.recv);
> + } else {
> + // drop packet();
Is this a commented function that's missing?
it's a comment. That means lwip does now own this packet and actual drop will be by U-Boot eth dev.
I will remove this comment to not confuse anybody.
> + LINK_STATS_INC(link.memerr);
> + LINK_STATS_INC(link.drop);
> + }
> +
> + return p;
> +}
> +
> +static int ethernetif_input(struct pbuf *p, struct netif *netif)
> +{
> + struct ethernetif *ethernetif;
> +
> + ethernetif = netif->state;
> +
> + /* move received packet into a new pbuf */
> + p = low_level_input(netif);
> +
> + /* if no packet could be read, silently ignore this */
> + if (p) {
> + /* pass all packets to ethernet_input, which decides what packets it supports */
> + if (netif->input(p, netif) != ERR_OK) {
> + LWIP_DEBUGF(NETIF_DEBUG, ("%s: IP input error\n", __func__));
> + pbuf_free(p);
> + p = NULL;
> + }
> + }
> +
> + return 0;
> +}
> +
> +static err_t low_level_output(struct netif *netif, struct pbuf *p)
> +{
> + int err;
> +
> + err = eth_send(p->payload, p->len);
> + if (err) {
> + log_err("eth_send error %d\n", err);
> + return ERR_ABRT;
> + }
> + return ERR_OK;
> +}
> +
> +err_t ulwip_if_init(struct netif *netif)
> +{
> + struct ulwip_if *uif;
> + struct ulwip *ulwip;
> +
> + uif = malloc(sizeof(struct ulwip_if));
> + if (!uif) {
> + log_err("uif: out of memory\n");
> + return ERR_MEM;
> + }
> + netif->state = uif;
> +
> + netif->name[0] = IFNAME0;
> + netif->name[1] = IFNAME1;
> +
> + netif->hwaddr_len = ETHARP_HWADDR_LEN;
> + string_to_enetaddr(env_get("ethaddr"), netif->hwaddr);
What if ethaddr is not set?
U-Boot set's it automatically. I think it will generate random one. But even if somehow it will not set
string_to_enetaddr() function just will do nothing.
> +#if defined(CONFIG_LWIP_LIB_DEBUG)
> + printf(" MAC: %02X:%02X:%02X:%02X:%02X:%02X\n",
> + netif->hwaddr[0], netif->hwaddr[1], netif->hwaddr[2],
> + netif->hwaddr[3], netif->hwaddr[4], netif->hwaddr[5]);
> +#endif
> +#if LWIP_IPV4
> + netif->output = etharp_output;
> +#endif
> +#if LWIP_IPV6
> + netif->output_ip6 = ethip6_output;
> +#endif
> +
> + netif->linkoutput = low_level_output;
> + netif->mtu = 1500;
> + netif->flags = NETIF_FLAG_BROADCAST | NETIF_FLAG_ETHARP | NETIF_FLAG_LINK_UP;
> +
> + ulwip = eth_lwip_priv(eth_get_dev());
> + ulwip->init_done = 1;
> + if (IS_ENABLED(CONFIG_LWIP_LIB_DEBUG)) {
> + log_info("Initialized LWIP stack\n");
> + }
> +
> + return ERR_OK;
> +}
> +
> +int ulwip_init(void)
> +{
> + ip4_addr_t ipaddr, netmask, gw;
> + struct netif *unetif;
> + struct ulwip *ulwip;
> + int ret;
> +
> + ret = eth_init();
> + if (ret) {
> + log_err("eth_init error %d\n", ret);
> + return ERR_IF;
> + }
> +
> + ulwip = eth_lwip_priv(eth_get_dev());
> + if (ulwip->init_done)
> + return CMD_RET_SUCCESS;
> +
> + unetif = malloc(sizeof(struct netif));
> + if (!unetif)
> + return ERR_MEM;
> + memset(unetif, 0, sizeof(struct netif));
> +
> + ip4_addr_set_zero(&gw);
> + ip4_addr_set_zero(&ipaddr);
> + ip4_addr_set_zero(&netmask);
> +
> + ipaddr_aton(env_get("ipaddr"), &ipaddr);
> + ipaddr_aton(env_get("ipaddr"), &netmask);
Here I missed "netmask" for &netmask.
> +
> + LWIP_PORT_INIT_NETMASK(&netmask);
> + if (IS_ENABLED(CONFIG_LWIP_LIB_DEBUG)) {
> + printf("Starting lwIP, IP: %s\n", ip4addr_ntoa(&ipaddr));
> + printf(" GW: %s\n", ip4addr_ntoa(&gw));
> + printf(" mask: %s\n", ip4addr_ntoa(&netmask));
log_info()
> + }
> +
> + if (!netif_add(unetif, &ipaddr, &netmask, &gw,
> + unetif, ulwip_if_init, ethernetif_input))
> + printf("err: netif_add failed!\n");
> +
> + netif_set_up(unetif);
> + netif_set_link_up(unetif);
> +#if LWIP_IPV6
> + netif_create_ip6_linklocal_address(unetif, 1);
> + printf(" IPv6: %s\n", ip6addr_ntoa(netif_ip6_addr(unetif, 0)));
> +#endif /* LWIP_IPV6 */
> +
> + return CMD_RET_SUCCESS;
> +}
> +
> +/* placeholder, not used now */
> +void ulwip_destroy(void)
> +{
> +}
> diff --git a/net/lwip/port/include/arch/cc.h b/net/lwip/port/include/arch/cc.h
> new file mode 100644
> index 0000000000..55f7787ce1
> --- /dev/null
> +++ b/net/lwip/port/include/arch/cc.h
> @@ -0,0 +1,39 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +
> +/*
> + * (C) Copyright 2023 Linaro Ltd. <maxim.uvarov@linaro.org>
> + */
> +
> +#ifndef LWIP_ARCH_CC_H
> +#define LWIP_ARCH_CC_H
> +
> +#include <linux/types.h>
> +#include <linux/kernel.h>
> +//#include <stdlib.h> /* getenv, atoi */
Please dont leave comments like that
> +#include <vsprintf.h>
> +
> +#define LWIP_ERRNO_INCLUDE <errno.h>
> +
> +#define LWIP_ERRNO_STDINCLUDE 1
> +#define LWIP_NO_UNISTD_H 1
> +#define LWIP_TIMEVAL_PRIVATE 1
Should those be defined in the LWIP config header instead?
I would keep it here, like ./net/lwip/lwip-external/contrib/ports/unix/port/include/arch/cc.h does.
Because LWIP config is more related to protocol features and settings and this header is about
porr for a specific platform.
> +
> +extern unsigned int lwip_port_rand(void);
This is like the forth time we go through this and it's a repeating
pattern. Why do we need this extern? Can't we just include the proper
header files?
> +#define LWIP_RAND() (lwip_port_rand())
This seems quite useless. Just use the function directly
#define LWIP_RAND() ((u32_t)rand())
> +
> +/* different handling for unit test, normally not needed */
> +#ifdef LWIP_NOASSERT_ON_ERROR
> +#define LWIP_ERROR(message, _expression_, handler) do { if (!(_expression_)) { \
> + handler; }} while (0)
> +#endif
> +
> +#define LWIP_DONT_PROVIDE_BYTEORDER_FUNCTIONS
> +
> +#define LWIP_PLATFORM_ASSERT(x) do {printf("Assertion \"%s\" failed at line %d in %s\n", \
> + x, __LINE__, __FILE__); } while (0)
> +
> +#define atoi(str) (int)dectoul(str, NULL)
> +
> +#define LWIP_ERR_T int
> +
> +#endif /* LWIP_ARCH_CC_H */
> diff --git a/net/lwip/port/include/arch/sys_arch.h b/net/lwip/port/include/arch/sys_arch.h
> new file mode 100644
> index 0000000000..92a8560d49
> --- /dev/null
> +++ b/net/lwip/port/include/arch/sys_arch.h
> @@ -0,0 +1,56 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +
> +/*
> + * (C) Copyright 2023 Linaro Ltd. <maxim.uvarov@linaro.org>
> + */
> +
> +#ifndef LWIP_ARCH_SYS_ARCH_H
> +#define LWIP_ARCH_SYS_ARCH_H
> +
> +#include "lwip/opt.h"
> +#include "lwip/arch.h"
> +#include "lwip/err.h"
> +
> +#define ERR_NEED_SCHED 123
> +
> +void sys_arch_msleep(u32_t delay_ms);
> +#define sys_msleep(ms) sys_arch_msleep(ms)
Dont redefine random functions here. U-Boot should already have all the
sleep functions you need
dropeed.
> +
> +#if SYS_LIGHTWEIGHT_PROT
Is this working? Can we define SYS_LIGHTWEIGHT_PROT?
dropeed.
> +typedef u32_t sys_prot_t;
> +#endif /* SYS_LIGHTWEIGHT_PROT */
> +
> +#include <errno.h>
> +
> +#define SYS_MBOX_NULL NULL
> +#define SYS_SEM_NULL NULL
> +
> +typedef u32_t sys_prot_t;
> +
> +typedef struct sys_sem *sys_sem_t;
> +#define sys_sem_valid(sem) (((sem) != NULL) && (*(sem) != NULL))
> +#define sys_sem_set_invalid(sem) do { if ((sem) != NULL) { *(sem) = NULL; }} while (0)
> +
> +/* let sys.h use binary semaphores for mutexes */
> +#define LWIP_COMPAT_MUTEX 1
> +#define LWIP_COMPAT_MUTEX_ALLOWED 1
> +
> +struct sys_mbox;
> +typedef struct sys_mbox *sys_mbox_t;
> +#define sys_mbox_valid(mbox) (((mbox) != NULL) && (*(mbox) != NULL))
> +#define sys_mbox_set_invalid(mbox) do { if ((mbox) != NULL) { *(mbox) = NULL; }} while (0)
All these macros seem unnecessary. Just assign types to NULL directly etc
ok.
> +
> +struct sys_thread;
> +typedef struct sys_thread *sys_thread_t;
> +
> +static inline u32_t sys_arch_sem_wait(sys_sem_t *sem, u32_t timeout)
> +{
> + return 0;
> +};
> +
> +static inline err_t sys_mbox_trypost(sys_mbox_t *mbox, void *msg)
> +{
> + return 0;
> +};
Are those really needed? Why do we just return 0?
Not needed, I think I planned to implement socket API later where it's needed. All these functions and macro can be defined to NULL with #define NO_SYS 1 in the lwip configuration.
This has to be an empty file.
> +
> +#endif /* LWIP_ARCH_SYS_ARCH_H */
> diff --git a/net/lwip/port/include/limits.h b/net/lwip/port/include/limits.h
> new file mode 100644
> index 0000000000..e69de29bb2
> diff --git a/net/lwip/port/sys-arch.c b/net/lwip/port/sys-arch.c
> new file mode 100644
> index 0000000000..609eeccf8c
> --- /dev/null
> +++ b/net/lwip/port/sys-arch.c
> @@ -0,0 +1,20 @@
> +// SPDX-License-Identifier: GPL-2.0
> +
> +/*
> + * (C) Copyright 2023 Linaro Ltd. <maxim.uvarov@linaro.org>
> + */
> +
> +#include <common.h>
> +#include <rand.h>
> +#include "lwip/opt.h"
> +
> +u32_t sys_now(void)
> +{
> + return get_timer(0);
> +}
> +
> +u32_t lwip_port_rand(void)
> +{
> + return (u32_t)rand();
I dont see why we cant use the U-Boot defined ones directly
changed.
> +}
> +
> --
> 2.30.2
>
Thanks
/Ilias