phw_sock_raw_mmaped not working in 0.6

[marcfir]

Hi Robert,
I tested the hw refactoring. RAW works fine, but RAWMMAPED seems to have a bug.
First, I think a hw_open call is missing in

libethercat/src/hw_sock_raw_mmaped.c

Lines 117 to 119 in e0be7c8

	}
	phw_sock_raw_mmaped->common.send = hw_device_sock_raw_mmaped_send;

But fixing this (https://github.com/marcfir/libethercat/tree/fix/rawmmaped) leads to a timeout. Looking with tcpdump shows that frames are outgoing. So I think the problem is in the recv function.

2024-09-02T08:59:18.832Z INFO  [libethercat_rs::base_types] Heap allocated for the ethercat master struct 68092064 bytes
2024-09-02T08:59:18.832Z INFO  [libethercat_rs::base_types] Heap allocated for hw_sock_raw_mmaped_t 13208 bytes
2024-09-02T08:59:18.832Z INFO  [libethercat_rs] HW_OPEN             : got page size 4096 bytes
2024-09-02T08:59:18.889Z INFO  [libethercat_rs] HW_OPEN             : binding raw socket to enp5s0
2024-09-02T08:59:18.889Z INFO  [libethercat_rs] HW_OPEN             : got mtu size 1500
2024-09-02T08:59:18.913Z INFO  [libethercat_rs] HW_OPEN             : Opend phw with size 13208
2024-09-02T08:59:18.939Z INFO  [libethercat_rs] ASYNC_LOOP          : async loop thread running
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         : libethercat version          : 0.5.1
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_SLAVES                 : 256
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_GROUPS                 : 8
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_PDLEN                  : 3036
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_MBX_ENTRIES            : 16
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_INIT_CMD_DATA          : 2048
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_SLAVE_FMMU             : 8
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_SLAVE_SM               : 8
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_DATAGRAMS              : 100
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_EEPROM_CAT_SM          : 8
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_EEPROM_CAT_FMMU        : 8
2024-09-02T08:59:18.940Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_EEPROM_CAT_PDO         : 128
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_EEPROM_CAT_PDO_ENTRIES : 32
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_EEPROM_CAT_STRINGS     : 128
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_EEPROM_CAT_DC          : 8
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_STRING_LEN             : 128
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_DATA                   : 4096
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_DS402_SUBDEVS          : 4
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_COE_EMERGENCIES        : 10
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         :   MAX_COE_EMERGENCY_MSG_LEN  : 32
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_OPEN         : Master struct needs 68092064 bytes
2024-09-02T08:59:18.941Z INFO  [libethercat_rs] MASTER_SET_STATE    : switching from EC_STATE_UNKNOWN to EC_STATE_INIT
2024-09-02T08:59:19.041Z ERROR [libethercat_rs] MASTER_TRANSCEIVE   : timeout on cmd 0x8, adr 0x1200000
2024-09-02T08:59:19.141Z ERROR [libethercat_rs] MASTER_TRANSCEIVE   : timeout on cmd 0x7, adr 0x0
2024-09-02T08:59:19.141Z ERROR [libethercat_rs] MASTER_SCAN         : broadcast read of slave types failed with 65600
thread 'main' panicked at libethercat-rs/examples/main.rs:37:43:
called `Result::unwrap()` on an `Err` value: StateNotReachable
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

tcpdump: verbose output suppressed, use -v[v]... for full protocol decode
listening on enp5s0, link-type EN10MB (Ethernet), snapshot length 262144 bytes
09:05:20.096764 00:30:64:0f:83:35 (oui Unknown) > Broadcast, ethertype Unknown (0x88a4), length 30: 
        0x0000:  1e10 0800 0000 2001 0200 0000 1100 0000  ................
09:05:20.096800 02:30:64:0f:83:35 (oui Unknown) > Broadcast, ethertype Unknown (0x88a4), length 60: 
        0x0000:  1e10 0800 0800 2001 0200 0000 1100 0800  ................
        0x0010:  0000 0000 0000 0000 0000 0000 0000 0000  ................
        0x0020:  0000 0000 0000 0000 0000 0000 0000       ..............
09:05:20.197009 00:30:64:0f:83:35 (oui Unknown) > Broadcast, ethertype Unknown (0x88a4), length 30: 
        0x0000:  1e10 0701 0000 0000 0200 0000 0000 0000  ................
09:05:20.197052 02:30:64:0f:83:35 (oui Unknown) > Broadcast, ethertype Unknown (0x88a4), length 60: 
        0x0000:  1e10 0701 0800 0000 0200 0000 1301 0800  ................
        0x0010:  0000 0000 0000 0000 0000 0000 0000 0000  ................
        0x0020:  0000 0000 0000 0000 0000 0000 0000       ..............

[robert-burger]

yes, you're right.
the creation of a receive thread is missing in the sock_raw_mmaped hw layer. i'll re-add this.

[robert-burger]

note: the open function pointer does not need to be addded to the hw_common struct as it needs to be called in prior to ec_open by the user. it does also take an arbitrary number of arguments depending on the needs of the hw layer so there's no way for libethercat call this...

[robert-burger]

should be fixed with 9108db1