Re: [PATCH bpf-next 1/1] bpf: fix WARNING in __bpf_prog_ret0_warn when jit failed

[Yonghong Song <yonghong.song@linux.dev>]

On 7/22/25 6:28 AM, Felix Fietkau wrote:
> Hi,
>
> On 26.05.25 15:33, KaFai Wan wrote:
>> syzkaller reported an issue:
>>
>> WARNING: CPU: 3 PID: 217 at kernel/bpf/core.c:2357 
>> __bpf_prog_ret0_warn+0xa/0x20 kernel/bpf/core.c:2357
>> Modules linked in:
>> CPU: 3 UID: 0 PID: 217 Comm: kworker/u32:6 Not tainted 
>> 6.15.0-rc4-syzkaller-00040-g8bac8898fe39 #0 PREEMPT(full)
>> Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 
>> 1.16.3-debian-1.16.3-2~bpo12+1 04/01/2014
>> Workqueue: ipv6_addrconf addrconf_dad_work
>> RIP: 0010:__bpf_prog_ret0_warn+0xa/0x20 kernel/bpf/core.c:2357
>> RSP: 0018:ffffc900031f6c18 EFLAGS: 00010293
>> RAX: 0000000000000000 RBX: ffffc9000006e000 RCX: 1ffff9200000dc06
>> RDX: ffff8880234ba440 RSI: ffffffff81ca6979 RDI: ffff888031e93040
>> RBP: ffffc900031f6cb8 R08: 0000000000000001 R09: 0000000000000000
>> R10: 0000000000000000 R11: 0000000000000000 R12: ffff88802b61e010
>> R13: ffff888031e93040 R14: 00000000000000a0 R15: ffff88802c3d4800
>> FS:  0000000000000000(0000) GS:ffff8880d6ce2000(0000) 
>> knlGS:0000000000000000
>> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> CR2: 000055557b6d2ca8 CR3: 000000002473e000 CR4: 0000000000352ef0
>> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>> Call Trace:
>>   <TASK>
>>   bpf_dispatcher_nop_func include/linux/bpf.h:1316 [inline]
>>   __bpf_prog_run include/linux/filter.h:718 [inline]
>>   bpf_prog_run include/linux/filter.h:725 [inline]
>>   cls_bpf_classify+0x74a/0x1110 net/sched/cls_bpf.c:105
>>   ...
>>
>> When creating bpf program, 'fp->jit_requested' depends on 
>> bpf_jit_enable.
>> Currently the value of bpf_jit_enable is available from 0 to 2, 0 
>> means use
>> interpreter and not jit, 1 and 2 means need to jit. When
>> CONFIG_BPF_JIT_ALWAYS_ON is enabled, bpf_jit_enable is permanently set
>> to 1, when it's not set or disabled, we can set bpf_jit_enable via proc.
>>
>> This issue is triggered because of CONFIG_BPF_JIT_ALWAYS_ON is not set
>> and bpf_jit_enable is set to 1, causing the arch to attempt JIT the 
>> prog,
>> but jit failed due to FAULT_INJECTION. As a result, incorrectly
>> treats the program as valid, when the program runs it calls
>> `__bpf_prog_ret0_warn` and triggers the WARN_ON_ONCE(1).
>>
>> Reported-by: syzbot+0903f6d7f285e41cdf10@syzkaller.appspotmail.com
>> Closes: 
>> https://lore.kernel.org/bpf/6816e34e.a70a0220.254cdc.002c.GAE@google.com
>> Fixes: fa9dd599b4da ("bpf: get rid of pure_initcall dependency to 
>> enable jits")
>> Signed-off-by: KaFai Wan <mannkafai@gmail.com>
>
> I think this patch may have caused a regression in configurations with 
> CONFIG_BPF_JIT_DEFAULT_ON=y when programs can't be JITed. Attaching 
> the program fails with error -ENOTSUPP.

Could you explain why there is an issue here?
CONFIG_BPF_JIT_DEFAULT_ON=y but prog cannot be jit'ed. So the end result is to return -ENOTSUPP.
It looks okay to me since the jit is required but jit failed, the only choice for the kernel
is to return an error.

>
> Please see https://github.com/openwrt/openwrt/issues/19405 for more 
> information.
>
> - Felix