Compiling of cake for gnuton build.

DJones · Jul 14, 2024

Hi @RMerlin

I’m trying to understand what indicates which model should cake be built into asuswrt-Merlin/gnuton.

I’ve been looking at why cake didn’t build into gnutons 388.7 release and have an open bug report on his git. As far as I can tell cake is set to be built as a module.

RT-AX58U Cake and FlexQoS not running due to Hardware acceleration runner being enabled · Issue #615 · gnuton/asuswrt-merlin.ng

Router Model Affected Models: RT-AX58U Firmware Version Affected 388.7_1 Is this bug present in upstream Merlin releases too? No Describe the bug Cake and FlexQoS doesn't work because Hardware acce...

github.com

However I’m curious about these profiles. Are they indicative of certain models (and this one was just absent from the profile) or is cake supposed to be built into every model and 388.7 was just a bad build for the newer supported device rt-ax58u v2. I haven’t dived further into the code then this because I’m not certain what I’m looking for.

release/src-rt-5.04axhnd.675x/targets/arch/PURE181.arch

Code:

<PUBLISH>
@@ special demark for maketargets.  All + and - lines after this line will modify profile.
-BUILD_DM_HYBRID=y
-BUILD_DM_LEGACY98=y
+BUILD_DM_PURE181=y
-BUILD_QUICKSETUP=y
+BUILD_QUICKSETUP is not set

RMerlin · Jul 14, 2024

DJones said:
I’m trying to understand what indicates which model should cake be built into asuswrt-Merlin/gnuton.

Build profiles are all in release/src-rt/target.mak.

DJones · Jul 14, 2024

I see references to BRCM which I assume is Broadcom. But nothing really related to cake as a module unless it umbrellas under something else.

Code:

export RT-AX58U_V2 := $(HND-96756_BASE)
export RT-AX58U_V2 += BUILD_NAME="RT-AX58U_V2" NVSIZE="128" DHDAP=n HND_WL=y DPSTA=n LACP=n WTFAST=n REPEATER=y IPV6SUPP=y HTTPS=y \
         ARM=y AUTODICT=y BBEXTRAS=y USBEXTRAS=y EBTABLES=y MEDIASRV=y MODEM=y USB_WAN_BACKUP=y PARENTAL2=y ACCEL_PPTPD=y PRINTER=y \
        WEBDAV=y SMARTSYNCBASE=y USB="USB" APP="network" PROXYSTA=y DNSMQ=y SHP=n BCMWL6=y BCMWL6A=y DISK_MONITOR=y BTN_WIFITOG=n \
        OPTIMIZE_XBOX=y ODMPID=y LED_BTN=n BCMSMP=y XHCI=y DUALWAN=y NEW_USER_LOW_RSSI=y OPENVPN=y TIMEMACHINE=y MDNS=y VPNC=y \
        JFFS2LOG=yNOTIFICATION_CENTER=y BWDPI=y DUMP_OOPS_MSG=n LINUX_MTD="64" DEBUGFS=y TEMPROOTFS=n SSH=y EMAIL=y FRS_FEEDBACK=y \
        SYSSTATE=y ROG=n STAINFO=y CLOUDCHECK=y NATNL_AICLOUD=y REBOOT_SCHEDULE=y MULTICASTIPTV=y QUAGGA=y WLCLMLOAD=n BCM_MUMIMO=y \
        LAN50="all" ATCOVER=y GETREALIP=y CFEZ=y TFAT=y NTFS="tuxera" HFS="tuxera" NEWSSID_REV2=y NEWSSID_REV4=y NEW_APP_ARM=y \
        VISUALIZATION=n BONDING=n BONDING_WAN=n NETOOL=y TRACEROUTE=y FORCE_AUTO_UPGRADE=y ALEXA=y IFTTT=y SW_HW_AUTH=y HD_SPINDOWN=y \
        ASPMD=n BCM_MEVENT=n BCMEVENTD=y BCM_APPEVENTD=n LETSENCRYPT=y VPN_FUSION=y JFFS_NVRAM=y NVRAM_ENCRYPT=y IPSEC=STRONGSWAN \
        IPSEC_SRVCLI_ONLY=SRV NATNL_AIHOME=y BCM_CEVENTD=y UTF8_SSID=y AMAS=y DWB=y DBLOG=y ETHOBD=y CFGSYNC_LOCSYNC=y CONNDIAG=y \
        NFCM=n DNSQUERY_INTERCEPT=n CRASHLOG=y WATCH_REINIT=n BW160M=y BRCM_HOSTAPD=y UUPLUGIN=n IPERF3=y FRS_LIVE_UPDATE=y ASUSCTRL=y \
        BCN_RPT=y BTM_11V=y INFO_EXAP=y AVBLCHAN=y SW_CTRL_ALLLED=y ASD=y AHS=y INTERNETCTRL=y GAME_MODE=n INSTANT_GUARD=y ACL96=y \
        GOOGLE_ASST=y COMFW=y WIREGUARD=y AMAS_ETHDETECT=y IPV6S46=y OCNVC=y
export RT-AX58U_V2 += TOR=y NTPD=y DNSPRIVACY=y DNSFILTER=y NTPD=y OPENSSL11=y NANO=y FORCE_AUTO_UPGRADE=n FRS_FEEDBACK=n
export RT-AX58U_V2 += UPNPIGD2=n VISUALIZATION=n IFTT=y LACP=n TUF_UI=n VPN_FUSION=n FRS_LIVE_UPDATE=n SNMPD=y

DJones · Jul 14, 2024

under release/src/router/rc/qos.c I’m seeing the same null pointers I was getting so maybe it’s not a case of cake not being installed.

Code:

#ifdef HND_ROUTER

int start_cake(void)
{
    unsigned int ibw, obw;
    FILE *f;
    char overheadstr[32];
    char ibwstr[32];
    char obwstr[32];
    char *mode;
    char nvnat[16];
    int nat;

    if((f = fopen("/etc/cake-qos.conf", "w")) == NULL) return -2;

    switch (nvram_get_int("qos_atm")) {
        case 0:
            mode = "";
            break;
        case 1:
            mode = "atm";
            break;
        case 2:
            mode = "ptm";
            break;
        default:
            mode = "";
    }

    ibw = strtoul(nvram_safe_get("qos_ibw"), NULL, 10);
    obw = strtoul(nvram_safe_get("qos_obw"), NULL, 10);

    if (ibw == 0)
        *ibwstr = '\0';
    else
        snprintf(ibwstr, sizeof(ibwstr), "bandwidth %dkbit", ibw);

    if (obw == 0)
        *obwstr = '\0';
    else
        snprintf(obwstr, sizeof(obwstr), "bandwidth %dkbit", obw);

    snprintf(overheadstr, sizeof(overheadstr), "overhead %d mpu %d", nvram_get_int("qos_overhead"), nvram_get_int("qos_mpu"));

    const char *wan_ifname = get_wan_ifname(wan_primary_ifunit());

    snprintf(nvnat, sizeof (nvnat), "wan%d_nat_x", wan_primary_ifunit());
    nat = nvram_get_int(nvnat);

    /* Config parameters */
    fprintf(f,
        "#!/bin/sh\n\n"
        "ULIF='%s'\n"
        "DLIF='%s'\n"
        "MIF='ifb4%s'\n"
        "ULBW='%s'\n"
        "DLBW='%s'\n"
        "OVERHEAD='%s'\n"
        "FRAMING='%s'\n"
        "ULPRIOQUEUE='diffserv3'\n"
        "DLPRIOQUEUE='besteffort'\n"
        "ULOPTIONS='%s dual-srchost'\n"
        "DLOPTIONS='%s wash dual-dsthost ingress'\n",

        wan_ifname,
        wan_ifname,
        wan_ifname,
        obwstr,
        ibwstr,
        overheadstr,
        mode,
        (nat ? "nat" : ""),
        (nat ? "nat" : "")
    );

    append_custom_config("cake-qos.conf",f);
    fclose(f);


    if((f = fopen(qosfn, "w")) == NULL) return -2;

    /* Stop/start rules */
    fprintf(f,
        "#!/bin/sh\n"
        "source /etc/cake-qos.conf\n\n"

        "case \"$1\" in\n"
        "start)\n"
        "# Upload\n"
        "\ttc qdisc add dev $ULIF root cake $ULPRIOQUEUE $ULBW $OVERHEAD $FRAMING $ULOPTIONS 2>/dev/null\n\n"

        "# Download\n"
        "\tip link add name $MIF type ifb 2>/dev/null\n"
        "\ttc qdisc add dev $DLIF handle ffff: ingress 2>/dev/null\n"
        "\ttc qdisc add dev $MIF root cake $DLPRIOQUEUE $DLBW $OVERHEAD $FRAMING $DLOPTIONS 2>/dev/null\n"
        "\tip link set $MIF up 2>/dev/null\n"
        "\ttc filter add dev $DLIF parent ffff: prio 10 matchall action mirred egress redirect dev $MIF 2>/dev/null\n\n"

        "\t;;\n"
        "stop)\n"
        "\ttc qdisc del dev $ULIF root 2>/dev/null\n"
        "\ttc qdisc del dev $DLIF ingress 2>/dev/null\n"
        "\ttc qdisc del dev $MIF root 2>/dev/null\n"
        "\tip link set $MIF down 2>/dev/null\n"
        "\tip link del dev $MIF 2>/dev/null\n"
        "\t;;\n"
        "*)\n"
        "esac\n");

    fclose(f);
    chmod("/etc/cake-qos.conf", 0755);
    chmod(qosfn, 0755);
    run_custom_script("qos-start", 120, "init", NULL);
    eval((char *)qosfn, "start");

    return 0;
}
#endif

Code:

@RT-AX58U-B260:/tmp/home/root# cat /tmp/qos
#!/bin/sh
source /etc/cake-qos.conf

case "$1" in
start)
# Upload
        tc qdisc add dev $ULIF root cake $ULPRIOQUEUE $ULBW $OVERHEAD $FRAMING $ULOPTIONS 2>/dev/null

# Download
        ip link add name $MIF type ifb 2>/dev/null
        tc qdisc add dev $DLIF handle ffff: ingress 2>/dev/null
        tc qdisc add dev $MIF root cake $DLPRIOQUEUE $DLBW $OVERHEAD $FRAMING $DLOPTIONS 2>/dev/null
        ip link set $MIF up 2>/dev/null
        tc filter add dev $DLIF parent ffff: prio 10 matchall action mirred egress redirect dev $MIF 2>/dev/null

        ;;
stop)
        tc qdisc del dev $ULIF root 2>/dev/null
        tc qdisc del dev $DLIF ingress 2>/dev/null
        tc qdisc del dev $MIF root 2>/dev/null
        ip link set $MIF down 2>/dev/null
        ip link del dev $MIF 2>/dev/null
        ;;
*)
esac

DJones · Jul 14, 2024

Something seems wrong with tc qdisc and I’m going to chalk it up to my initial concerns that hw acceleration runner would not turn off in 388.7.. will test 388.8 on gnuton tomorrow.

Code:

@RT-AX58U-B260:/tmp/home/root# tc qdisc
qdisc noqueue 0: dev lo root refcnt 2
qdisc noqueue 0: dev archer root refcnt 2
qdisc pfifo_fast 0: dev eth0 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc ingress ffff: dev eth0 parent ffff:fff1 ----------------
qdisc pfifo_fast 0: dev eth1 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev spu_us_dummy root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev spu_ds_dummy root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth2 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth3 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc noqueue 0: dev br0 root refcnt 2
qdisc noqueue 0: dev wgs1 root refcnt 2
qdisc pfifo_fast 0: dev ifb4eth0 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1

@RT-AX58U-B260:/tmp/home/root# tc qdisc add dev eth0 root cake diffserv3 bandwidth 2355kbit overhead 18 mpu 64 atm nat dual-srchost
RTNETLINK answers: No such file or directory

RMerlin · Jul 15, 2024

DJones said:
I see references to BRCM which I assume is Broadcom. But nothing really related to cake as a module unless it umbrellas under something else.

I think I simply enable it in the kernel config for every HND platform rather than enable it on a model basis.

You should check the content of /proc/config.gz on the router to determine if CAKE is enabled or not in that kernel.

You are talking about Gnuton and RT-AX58U, so I take it you mean RT-AX58U V2. I don't know if that device uses the same CPU as the V1, if not then its kernel config file will need the Asuswrt-Merlin specific kernel options applied. Check the config files within kernel/linux-4.1/ that's specific to that model's CPU, and make sure it has the same changes as those I applied to config_base.6a.6750 for the V1 version of that router. See the commit log for that file to track down all the changes I've done over the original config file.

DJones · Jul 15, 2024

RMerlin said:
I think I simply enable it in the kernel config for every HND platform rather than enable it on a model basis.

You should check the content of /proc/config.gz on the router to determine if CAKE is enabled or not in that kernel.

You are talking about Gnuton and RT-AX58U, so I take it you mean RT-AX58U V2. I don't know if that device uses the same CPU as the V1, if not then its kernel config file will need the Asuswrt-Merlin specific kernel options applied. Check the config files within kernel/linux-4.1/ that's specific to that model's CPU, and make sure it has the same changes as those I applied to config_base.6a.6750 for the V1 version of that router. See the commit log for that file to track down all the changes I've done over the original config file.

I’ll give /proc/config.gz a look when I have some free time later today. But the rt-ax58u v2 uses a BCM6756 a 4 core compared to the v1 which is a 3 core cpu BCM6750. So it’s possible that the configuration may need to be changed.

ASUS RT-AX58U v2 - TechInfoDepot

techinfodepot.shoutwiki.com

ASUS RT-AX58U - TechInfoDepot

techinfodepot.shoutwiki.com

DJones · Jul 15, 2024

Here’s the diff of both config_base.6a.67XX files what jumps out particularly is codel is not set on v2.

Code:

# CONFIG_NET_SCH_CODEL is not set

https://www.diffchecker.com/eyFX3IJn/

DJones · Jul 15, 2024

I just checked /proc/config.gz but it isn't in human readable format. edit: ahh it's compressed

config - Pastebin.com

Pastebin.com is the number one paste tool since 2002. Pastebin is a website where you can store text online for a set period of time.

pastebin.com

Code:

# CONFIG_NET_SCH_CAKE is not set

which is true at the time of checking. I'm not sure if that changes automatically at build of the firmware or on the fly.

Edit2: it's not set. That answers that I guess.

Edit3: Also not set in his beta of 388.8

DJones · Jul 15, 2024

Yet to test the build, but Fixed 875b94d

Thank you @RMerlin for pointing me in the right direction.

DJones · Jul 16, 2024

For my own sanity is their anything that would prevent Hw acceleration archer/runner from being turned off even if

Code:

nvram set runner_disable_force=1
nvram commit

even tried

Code:

nvram set archer_disable_force=1
nvram commit

The one glaring problem to me has been archer won't turn off, even if cake is enabled. Cake should turn it off automatically I assume. Now I've looked through the code, but I can't see anything apparent.

This is a error I did get in system logs.

DJones · Jul 18, 2024

might need to add RTCONFIG_HND_ROUTER_AX_6756 to these ifs.

asuswrt-merlin.ng/release/src/router/rc/sysdeps/init-broadcom.c
Lines 12156 to 12166 in 875b94d

#if defined(RTCONFIG_HND_ROUTER_AX_675X)
/* apply archer instead of runner */
eval("fc", "config", "--hw-accel", "0");
#else
eval("runner", "disable");
#endif
else if (!bootup)
#if defined(RTCONFIG_HND_ROUTER_AX_675X)
/* apply archer instead of runner */
eval("fc", "config", "--hw-accel", "1");
eval("archerctl", "sysport_tm", "disable");

As far as I can tell init-broadcom.c is a template file.

release/src-rt-5.04axhnd.675x/router-sysdep.rt-ax58u_v2. Contains archerctl. The actual files might be compiled as a .o object file which is not human readable.

is likely the real location for the init-broadcom

release/src/router/rc/prebuild/RT-AX58U_V2/broadcom.o

Probably contains (or doesn’t) the if s calling on the main archerctl

Anyway to convert .o files back into human readable format? I can look at it as hex/binary, but might as well be Latin.

RMerlin · Jul 18, 2024

DJones said:
Anyway to convert .o files back into human readable format? I can look at it as hex/binary, but might as well be Latin.

Unless you understand ARM ASM code, there`s nothing for you to look at if you disassembled the code.

Code:

$ arm-linux-objdump -d broadcom.o | more

broadcom.o:     file format elf32-littlearm


Disassembly of section .text.config_eth_port_shaper:

00000000 <config_eth_port_shaper>:
   0:    e92d42f0     push    {r4, r5, r6, r7, r9, lr}
   4:    e24dd068     sub    sp, sp, #104    ; 0x68
   8:    e28d4004     add    r4, sp, #4
   c:    e28d5014     add    r5, sp, #20
  10:    e1a09000     mov    r9, r0
  14:    e1a06001     mov    r6, r1
  18:    e3a02010     mov    r2, #16
  1c:    e3a01000     mov    r1, #0
  20:    e1a00004     mov    r0, r4
  24:    e28d7024     add    r7, sp, #36    ; 0x24
  28:    ebfffffe     bl    0 <memset>
  2c:    e3a02010     mov    r2, #16
  30:    e3a01000     mov    r1, #0
  34:    e1a00005     mov    r0, r5
  38:    ebfffffe     bl    0 <memset>
  3c:    e3a02010     mov    r2, #16
  40:    e3a01000     mov    r1, #0
  44:    e1a00007     mov    r0, r7
  48:    ebfffffe     bl    0 <memset>
 
...

And even if you did, you can't change anything about it either.

RMerlin · Jul 18, 2024

Holy crap... I can't vouch for the accuracy because I can't be bothered to review it (my ASM coding days are long behind me), but ChatGPT can apparently rewrite ASM code into C Code

DJones · Jul 18, 2024

RMerlin said:

Unless you understand ARM ASM code, there`s nothing for you to look at if you disassembled the code.

Code:

$ arm-linux-objdump -d broadcom.o | more

broadcom.o:     file format elf32-littlearm


Disassembly of section .text.config_eth_port_shaper:

00000000 <config_eth_port_shaper>:
   0:    e92d42f0     push    {r4, r5, r6, r7, r9, lr}
   4:    e24dd068     sub    sp, sp, #104    ; 0x68
   8:    e28d4004     add    r4, sp, #4
   c:    e28d5014     add    r5, sp, #20
  10:    e1a09000     mov    r9, r0
  14:    e1a06001     mov    r6, r1
  18:    e3a02010     mov    r2, #16
  1c:    e3a01000     mov    r1, #0
  20:    e1a00004     mov    r0, r4
  24:    e28d7024     add    r7, sp, #36    ; 0x24
  28:    ebfffffe     bl    0 <memset>
  2c:    e3a02010     mov    r2, #16
  30:    e3a01000     mov    r1, #0
  34:    e1a00005     mov    r0, r5
  38:    ebfffffe     bl    0 <memset>
  3c:    e3a02010     mov    r2, #16
  40:    e3a01000     mov    r1, #0
  44:    e1a00007     mov    r0, r7
  48:    ebfffffe     bl    0 <memset>
 
...

And even if you did, you can't change anything about it either.

Fair enough. Thanks for the input. I’m hardly the expert on any of this; I can work my way through code well enough to try and understand maybe something is missing and differentiate the code between versions. But just aimlessly searching since I haven’t worked on this like you or gnuton have to know where things might be.

Anyways going to let gnuton do his thing, too many cooks so to speak.

Tvlz · Jul 19, 2024

DJones said:
Here’s the diff of both config_base.6a.67XX files what jumps out particularly is codel is not set on v2.

https://www.diffchecker.com/eyFX3IJn/

You are still missing kernel config options(link), your diff is comparing the wrong files.
The ax58u v1 builds from "release/src-rt-5.02axhnd.675x"
The ax58u v2 builds from "release/src-rt-5.04axhnd.675x"

DJones · Jul 19, 2024

Tvlz said:
You are still missing kernel config options(link), your diff is comparing the wrong files.
The ax58u v1 builds from "release/src-rt-5.02axhnd.675x"
The ax58u v2 builds from "release/src-rt-5.04axhnd.675x"

Good catch!

DJones · Jul 19, 2024

Tvlz said:
You are still missing kernel config options(link), your diff is comparing the wrong files.
The ax58u v1 builds from "release/src-rt-5.02axhnd.675x"
The ax58u v2 builds from "release/src-rt-5.04axhnd.675x"

Yeah I’m getting mixed up jumping between rmerlins repositories and gnutons my intent was to differentiate between v1 and v2 but of different hardware revisions. 5.04 of rmerlin of cpu 6750(v1) and gnuton 6756(v2)

Edit: the more I look the more confused I get shouldn’t look at this on my phone. Gnuton has builds for 5.02 and 5.04 for the rt-ax58u v2 which is likely why I did the diff that way, but rmerl only has 5.02 of the rt-ax58u v1

Tvlz · Jul 20, 2024

DJones said:
my intent was to differentiate between v1 and v2 but of different hardware revisions. 5.04 of rmerlin of cpu 6750(v1) and gnuton 6756(v2)

To compare only the 5.04 builds you would need to diff "config_base.6a.4912" vs "config_base.6a.6756"(rt-ax58u v2) as rmerlin only builds 5.04 for routers using "config_base.6a.4912" so the other "config_base.6a.*" lack rmerlin's config options.

DJones · Jul 20, 2024

Tvlz said:
To compare only the 5.04 builds you would need to diff "config_base.6a.4912" vs "config_base.6a.6756"(rt-ax58u v2) as rmerlin only builds 5.04 for routers using "config_base.6a.4912" so the other "config_base.6a.*" lack rmerlin's config options.

Thanks, I’ve relayed this information to the bug report. Appreciate the better eyes on this. Lol

Thread starter	Title	Forum	Replies	Date
H	Need instructions on compiling Merlin rom	Asuswrt-Merlin	5	Mar 29, 2024
S	Issues compiling Realtek RTL8156 (r8152) driver for ASUS RT-AX86U	Asuswrt-Merlin	2	Mar 24, 2024
P	Error compiling for AX58U: cannot find crt1.o crti.o crtbegin.o	Asuswrt-Merlin	4	Jan 17, 2024
W	Compiling and toolchain questions for RT-AX86U Pro	Asuswrt-Merlin	6	Jan 10, 2024
	AX86U Merlin and Cake QoS	Asuswrt-Merlin	34	Aug 27, 2024
A	How do I configure cake QoS for FTTH?	Asuswrt-Merlin	4	Jul 25, 2024
D	Solved Gnuton RT-AX58U v2 issues with cake and skynet.	Asuswrt-Merlin	11	May 31, 2024
A	Sudden increase in bufferbloat with Cake QoS when nothing's changed?	Asuswrt-Merlin	3	May 7, 2024
T	Asus RT-AX86U vs RT-AX68U (with Merlin cake SQM)	Asuswrt-Merlin	9	Apr 13, 2024
	QOS Cake with merlin and MTU - slow download speeds / inconsistent mtu fragmentation results	Asuswrt-Merlin	14	Feb 18, 2024

Compiling of cake for gnuton build.

Very Senior Member

Asuswrt-Merlin dev

Very Senior Member

Very Senior Member

Very Senior Member

Asuswrt-Merlin dev

Very Senior Member

Very Senior Member

Very Senior Member

Very Senior Member

Very Senior Member

Very Senior Member

Asuswrt-Merlin dev

Asuswrt-Merlin dev

Very Senior Member

Occasional Visitor

Very Senior Member

Very Senior Member

Occasional Visitor

Very Senior Member

Similar threads

Similar threads

Support SNBForums w/ Amazon

Sign Up For SNBForums Daily Digest