The rpmsg_ioctl() function will first obtain the read lock before
traversing g_rpmsg, then the rpmsg_dev_destory function will try
to obtain the write lock.
At this time, rpmsg_ioctl does not release the read lock, resulting
in a deadlock.
To solve this problem, add a standardalone list for the rptun instance
to avoid hold the rpmsg lock when execute rptun_poweroff()
Backtrace:
Thread 14 nsh_main
0 0x4412b9da in up_switch_context (tcb=0xeac03380, rtcb=0xe7f02480) at sim/sim_switchcontext.c:95
1 0x43fbe5e4 in nxsem_wait_slow (sem=0x5179d570 <g_rpmsg_lock+16>) at semaphore/sem_wait.c:170
2 0x43fbeacf in nxsem_wait (sem=0x5179d570 <g_rpmsg_lock+16>) at semaphore/sem_wait.c:272
3 0x44671a26 in down_write (rwsem=0x5179d560 <g_rpmsg_lock>) at semaphore/sem_rw.c:248
4 0x4401e255 in rpmsg_device_destory (rpmsg=0xe7b03280) at rpmsg/rpmsg.c:469
5 0x4402f5fe in rptun_dev_stop (rproc=0xe7b033e4, stop_ns=true) at rptun/rptun.c:973
6 0x4402aa28 in rptun_ioctl (rpmsg=0xe7b03280, cmd=11109, arg=0) at rptun/rptun.c:661
7 0x4401ba1e in rpmsg_dev_ioctl_ (rpmsg=0xe7b03280, cmd=11109, arg=0) at rpmsg/rpmsg.c:137
8 0x4401ee88 in rpmsg_ioctl (cpuname=0x0, cmd=11109, arg=0) at rpmsg/rpmsg.c:562
9 0x44030cb0 in rptun_poweroff (cpuname=0x0) at rptun/rptun.c:1183
10 0x43f27be2 in board_power_off (status=0) at sim/sim_head.c:211
11 0x446fd933 in boardctl (cmd=65283, arg=0) at boardctl.c:401
12 0x44313b95 in cmd_poweroff (vtbl=0xe7600e80, argc=1, argv=0xe7824350) at nsh_syscmds.c:368
13 0x442d148c in nsh_command (vtbl=0xe7600e80, argc=1, argv=0xe7824350) at nsh_command.c:1275
14 0x442f65f5 in nsh_execute (vtbl=0xe7600e80, argc=1, argv=0xe7824350, param=0xe78242b0) at nsh_parse.c:716
15 0x44304fb4 in nsh_parse_command (vtbl=0xe7600e80, cmdline=0xe7601338 "poweroff") at nsh_parse.c:2809
16 0x44305a50 in nsh_parse (vtbl=0xe7600e80, cmdline=0xe7601338 "poweroff") at nsh_parse.c:2919
17 0x442bdc91 in nsh_session (pstate=0xe7600e80, login=1, argc=1, argv=0xe7814870) at nsh_session.c:246
18 0x442bbbe5 in nsh_consolemain (argc=1, argv=0xe7814870) at nsh_consolemain.c:75
19 0x4422b4bb in nsh_main (argc=1, argv=0xe7814870) at nsh_main.c:74
20 0x440ee098 in nxtask_startup (entrypt=0x4422b39a <nsh_main>, argc=1, argv=0xe7814870) at sched/task_startup.c:72
21 0x43fa1d6e in nxtask_start () at task/task_start.c:116
22 0x4412982c in pre_start () at sim/sim_initialstate.c:53
23 0x00000000 in ?? ()
Signed-off-by: wangshaoxin <wangshaoxin@xiaomi.com>
When execute 'rptun stop /dev/rptun/proxy' in the nsh,
rpmsg_deinit_vdev(&priv->rvdev) will set rvdev->vdev = NULL,
then the remoteproc_remove_virtio(priv->rvdev.vdev) input is NULL,
which will cause crash, the backtrace as follow, so we save the
rvdev->vdev pointer on vdev to solve this problem.
(gdb) bt
0 _assert (filename=0x4002aecc <nxrmutex_destroy+8> "\005D\220\v", linenum=1038, msg=0x400a26a8 "tx_vq", regs=0x400a2768) at misc/assert.c:827
1 0x4002a1db in __assert (filename=0x400a2204 "open-amp/lib/remoteproc/remoteproc.c", linenum=1038, msg=0x0) at assert/lib_assert.c:38
2 0x4006fd83 in remoteproc_remove_virtio (rproc=0xf3ea1880, vdev=0x0) at open-amp/lib/remoteproc/remoteproc.c:1038
3 0x4001b01d in rptun_dev_stop (rproc=0xf3ea1880, stop_ns=true) at rptun/rptun.c:979
4 0x4001a45e in rptun_ioctl (rpmsg=0xf3ea16d0, cmd=11109, arg=0) at rptun/rptun.c:661
5 0x40017246 in rpmsg_dev_ioctl_ (rpmsg=0xf3ea16d0, cmd=11109, arg=0) at rpmsg/rpmsg.c:138
6 0x4001728e in rpmsg_dev_ioctl (filep=0xf3d70a14, cmd=11109, arg=0) at rpmsg/rpmsg.c:153
7 0x4005ef51 in file_vioctl (filep=0xf3d70a14, req=11109, ap=0xf3d80f2c "\352\r\004@F\023\004@\230\020\330\363\353;\354\363(7\354", <incomplete sequence \363>)
at vfs/fs_ioctl.c:70
8 0x4005f2ff in ioctl (fd=3, req=11109) at vfs/fs_ioctl.c:294
9 0x40041049 in cmd_rpmsg_once (vtbl=0xf3ec3728, path=0xf3ec3beb "/dev/rptun/proxy", argv=0xf3d81098, rpmsg_cb=0x40041346 <cmd_rptun_cb>) at nsh_syscmds.c:616
10 0x400415c8 in cmd_rptun (vtbl=0xf3ec3728, argc=3, argv=0xf3d81098) at nsh_syscmds.c:766
11 0x40036024 in nsh_command (vtbl=0xf3ec3728, argc=3, argv=0xf3d81098) at nsh_command.c:1275
12 0x4003d044 in nsh_execute (vtbl=0xf3ec3728, argc=3, argv=0xf3d81098, param=0xf3d81078) at nsh_parse.c:716
13 0x4003efcc in nsh_parse_command (vtbl=0xf3ec3728, cmdline=0xf3ec3be0 "rptun") at nsh_parse.c:2809
14 0x4003f129 in nsh_parse (vtbl=0xf3ec3728, cmdline=0xf3ec3be0 "rptun") at nsh_parse.c:2919
15 0x40033ca4 in nsh_session (pstate=0xf3ec3728, login=1, argc=1, argv=0xf3d70ae0) at nsh_session.c:246
16 0x400338ea in nsh_consolemain (argc=1, argv=0xf3d70ae0) at nsh_consolemain.c:75
17 0x40033845 in nsh_main (argc=1, argv=0xf3d70ae0) at nsh_main.c:74
18 0x4002bc9f in nxtask_startup (entrypt=0x400337d0 <nsh_main>, argc=1, argv=0xf3d70ae0) at sched/task_startup.c:72
19 0x400077cc in nxtask_start () at task/task_start.c:116
20 0x40031d2f in pre_start () at sim/sim_initialstate.c:53
21 0x00000000 in ?? ()
Signed-off-by: mazhuang <mazhuang@xiaomi.com>
return -EAGAIN will cause rpmsg_virtio_get_tx_payload_buffer() in
OpenAMP break immediately, but should return to the sleep() logic
to wait buffer is return.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
Signed-off-by: Yongrong Wang <wangyongrong@xiaomi.com>
fix build break:
ld: drivers/libdrivers.a(rptun_ivshmem.c.obj): in function `rptun_ivshmem_get_resource':
nuttx/drivers/rptun/rptun_ivshmem.c:215:(.text.rptun_ivshmem_get_resource+0xfc): undefined reference to `simple_addrenv_initialize'
ld: openamp/libmetal/lib/libmetal.a(io.c.obj): in function `metal_io_phys_to_offset_':
nuttx/openamp/libmetal/lib/system/nuttx/io.c:112:(.text.metal_io_phys_to_offset_+0xc): undefined reference to `up_addrenv_pa_to_va'
Signed-off-by: chao an <anchao.archer@bytedance.com>
Most tools used for compliance and SBOM generation use SPDX identifiers
This change brings us a step closer to an easy SBOM generation.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
1. add IS_ALIGNED() definitions for NuttX;
2. replace all the ALIGN_UP() and ALIGN_DOWN() to use common
align implementation;
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
notify_wait has been remove in remoteproc ops and virtio_dispatch and
has been moved to the struct rpmsg_virtio_device in new OpenAMP,
so change the rptun too,
Signed-off-by: wangyongrong <wangyongrong@xiaomi.com>
1. Add cmake support for rptun_serure;
2. Move rptun_secure.c before rptun_ivshmem.c in Make.defs to follow
the order in Kconfig.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
Master can send restart command to slave to reboot the slave core
Signed-off-by: mazhuang <mazhuang@xiaomi.com>
Signed-off-by: Yongrong Wang <wangyongrong@xiaomi.com>
Because locate the command at the end the resource table is unfriendly
when we want to support multi virtio devices instead only one virtio
rpmsg device.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
Add more common command for rptun and rpmsg_virtio frameworks,
also modify the rptun and rpmsg_virtio driver to use the common
commands.
Signed-off-by: Yongrong Wang <wangyongrong@xiaomi.com>
Move the panic logic in common places, later we can move more logic to
the framework instead of having the drivers implement it repeatedly.
Signed-off-by: Yongrong Wang <wangyongrong@xiaomi.com>
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
headrx is very convient to check weather current core miss interrupt
by comparing the headrx with the rx vring avail.idx for slave side or
rx vring used.idx for master side.
So move headrx out of the CONFIG_RPTUN_PM range.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
rptun_dump related code has been moved to rptun.c from rptun_dump.c,
but file rptun_dump.c is not deleted in PR:
https://github.com/apache/nuttx/pull/11712
So delete this file.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
Store the rx virtqueue idx to the local headrx index, and only
process the data when the rx virtqueue has data to avoid access
the ram in low power mode.
Signed-off-by: ligd <liguiding1@xiaomi.com>
rptun secure is a rptun driver used for the rpmsg communication
between (Non-Secure) REE and (Secure) TEE environments.
Signed-off-by: yanghuatao <yanghuatao@xiaomi.com>
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
Add get_local_cpuname to the rpmsg framework ops to support communicate
with the same remote core with multi rpmsg transport.
Some rpmsg services will send local cpu name to remote core and then let
remote core to connect local core by using this cpu name, when there are
multi rpmsg channels with same remote core, the remote core may connect
to incorrect core, so use the error rpmsg channel.
For example, there are two rpmsg channels between ap and audio:
ap core audio core
[ap1] <-- rpmsg virtio1 --> [audio1]
[ap2] <-- rpmsg virtio2 --> [audio2]
When we want to use the rpmsg virtio1 to communicate, ap core may send
local cpuname "ap2" to audio, so the audio core use remote cpu "ap2" to
connect with ap, and resulting in the use of incorrect rpmsg channel.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
Rptun is a rptun fremework driver base on the Inter-Virtual Machine
Share Memory Device.
With this driver, two NuttX runs in different QEMU can communicate
with each other by the rpmsg api.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
In the previous patch, we moved rptun_panic and rptun_dump_all in rptun to rpmsg.
In order to ensure CI pass, this two API in rptun is not removed. So we remove rptun_panic and rptun_dump_all in this patch.
Signed-off-by: wangyongrong <wangyongrong@xiaomi.com>
It should be possible to use ping without rptun being enabled, so striped rptun_ping out of rptun and rename to rpmsg_ping.
Signed-off-by: wangyongrong <wangyongrong@xiaomi.com>
