[Devel] [PATCH vz10 07/24] blk-cbt: don't WARN on a user-supplied ABI version mismatch

[Konstantin Khorenko]

On 7/6/26 14:11, Andrey Zhadchenko wrote:
> I don't like that. Customers do not run with panic_on_warn. If this 
> fails in our test environment, that's actually great (it means something 
> went very wrong). pr_warn_ratelimited is worse than WARN_ONCE regarding 
> intentional spamming.

i agree, i will drop this patch.
It could definitely be useful in case there was WARN() - to change it to WARN_ONCE()
because this is triggerable from inside a Container, i have just checked that.

But as this is just a single WARN_ONCE, it's not a problem.

On the other hand mainstream fights with such user triggerable WARN_ONCE messages as well, like

commit 251a8fe1b9aedccd298b77bc28426d564c5a923f
Author: Masami Hiramatsu (Google) <mhiramat at kernel.org>
Date:   Thu Jun 25 08:34:46 2026 +0900

    tracing/probes: Remove WARN_ON_ONCE from parse_btf_arg

    Sashiko found that user can cause this WARN_ON_ONCE() easily
    with adding a kprobe event based on a raw address with BTF
    parameter.

    Since this is not an unexpected condition, remove the
    WARN_ON_ONCE().

    Link: https://lore.kernel.org/all/178177265367.2059927.13789953014706792126.stgit@mhiramat.tok.corp.google.com/

    Link: https://sashiko.dev/#/patchset/178165816303.269421.7302603996990753309.stgit%40devnote2

    Reported-by: Sashiko <sashiko-bot at kernel.org>
    Fixes: b576e09701c7 ("tracing/probes: Support function parameters if BTF is available")
    Signed-off-by: Masami Hiramatsu (Google) <mhiramat at kernel.org>

diff --git a/kernel/trace/trace_probe.c b/kernel/trace/trace_probe.c
index fd1caa1f97233..98532c503d028 100644
--- a/kernel/trace/trace_probe.c
+++ b/kernel/trace/trace_probe.c
@@ -678,7 +678,7 @@ static int parse_btf_arg(char *varname,
        int i, is_ptr, ret;
        u32 tid;

-       if (WARN_ON_ONCE(!ctx->funcname && !(ctx->flags & TPARG_FL_TEVENT)))
+       if (!ctx->funcname && !(ctx->flags & TPARG_FL_TEVENT))
                return -EINVAL;

        is_ptr = split_next_field(varname, &field, ctx);

commit 40c88c429a598006f91ad7a2b89856cd50b3a008
Author: Andrii Nakryiko <andrii at kernel.org>
Date:   Tue May 16 11:04:09 2023 -0700

    bpf: drop unnecessary user-triggerable WARN_ONCE in verifierl log

    [ Upstream commit cff36398bd4c7d322d424433db437f3c3391c491 ]

    It's trivial for user to trigger "verifier log line truncated" warning,
    as verifier has a fixed-sized buffer of 1024 bytes (as of now), and there are at
    least two pieces of user-provided information that can be output through
    this buffer, and both can be arbitrarily sized by user:
      - BTF names;
      - BTF.ext source code lines strings.

    Verifier log buffer should be properly sized for typical verifier state
    output. But it's sort-of expected that this buffer won't be long enough
    in some circumstances. So let's drop the check. In any case code will
    work correctly, at worst truncating a part of a single line output.

    Reported-by: syzbot+8b2a08dfbd25fd933d75 at syzkaller.appspotmail.com
    Signed-off-by: Andrii Nakryiko <andrii at kernel.org>
    Link: https://lore.kernel.org/r/20230516180409.3549088-1-andrii@kernel.org
    Signed-off-by: Alexei Starovoitov <ast at kernel.org>
    Signed-off-by: Sasha Levin <sashal at kernel.org>

diff --git a/kernel/bpf/log.c b/kernel/bpf/log.c
index 920061e38d2e1..cd1b7113fbfd0 100644
--- a/kernel/bpf/log.c
+++ b/kernel/bpf/log.c
@@ -22,9 +22,6 @@ void bpf_verifier_vlog(struct bpf_verifier_log *log, const char *fmt,

        n = vscnprintf(log->kbuf, BPF_VERIFIER_TMP_LOG_SIZE, fmt, args);

-       WARN_ONCE(n >= BPF_VERIFIER_TMP_LOG_SIZE - 1,
-                 "verifier log line truncated - local buffer too short\n");
-
        if (log->level == BPF_LOG_KERNEL) {
                bool newline = n > 0 && log->kbuf[n - 1] == '\n';


> On 7/6/26 12:59, Konstantin Khorenko wrote:
>> blk_cbt_ioctl() reads abi_version straight from the ioctl argument and
>> WARN_ONCE()s if it does not match CBT_ABI_VERSION. The value is fully
>> userspace-controlled, so any process issuing a BLKCBT* ioctl with a
>> stale/newer struct taints the kernel, dumps a backtrace, and can panic a
>> host that runs with panic_on_warn. Downgrade to pr_warn_ratelimited();
>> the -EOPNOTSUPP return is the actual contract.
>>
>> Fixes: 6e42f62a2c88 ("block/blk-cbt: introduce ABI versioning")
>> Feature: cbt: changed block tracking (for backup)
>> https://virtuozzo.atlassian.net/browse/VSTOR-137234
>> Signed-off-by: Konstantin Khorenko <khorenko at virtuozzo.com>
>> ---
>>   block/blk-cbt.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/block/blk-cbt.c b/block/blk-cbt.c
>> index 90219f58f1ae..91a888770411 100644
>> --- a/block/blk-cbt.c
>> +++ b/block/blk-cbt.c
>> @@ -1093,8 +1093,8 @@ int blk_cbt_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
>>   		return -EFAULT;
>>   
>>   	if (abi_version != CBT_ABI_VERSION) {
>> -		WARN_ONCE(1, "blk-cbt ABI mimatch: kernel has %d, userspace uses %d",
>> -			  CBT_ABI_VERSION, abi_version);
>> +		pr_warn_ratelimited("blk-cbt: ABI mismatch: kernel has %d, userspace uses %d\n",
>> +				    CBT_ABI_VERSION, abi_version);
>>   		return -EOPNOTSUPP;
>>   	}
>>   
>