[PATCH v2 bpf-next 2/3] net: Improvement for bpf_sysctl_set_new_value

[Raman Shukhau <ramasha@meta.com>]

When bpf_sysctl_set_new_value is called in cgroup/sysctl handler, updated
"new_len" is provided back to proc_sys_call_handler. But
proc_sys_call_handler	expects this value NOT to include \0 symbol, e.g.
if user do:

open("/proc/sys/net/ipv4/ip_local_reserved_ports", ...)
write(fd, "11111", sizeof("22222"))

or

echo -n "11111" > /proc/sys/net/ipv4/ip_local_reserved_ports

or

sysctl -w	net.ipv4.ip_local_reserved_ports=11111

proc_sys_call_handler receives count equal to `5`. if BPF handler code
doesn't account for that, new value will be rejected by
proc_sys_call_handler with EINVAL error.

To make behavior consistent for bpf_sysctl_set_new_value, this change
adjust `new_len` with `-1`, if `\0` passed as last character.
Alternatively, using `sizeof("11111") - 1` in BPF handler should work,
but it might not be obvious and spark confusion.

Signed-off-by: Raman Shukhau <ramasha@meta.com>
---
 kernel/bpf/cgroup.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index bfc36e7ca6f6..23736aed1b53 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -1742,7 +1742,10 @@ int __cgroup_bpf_run_filter_sysctl(struct ctl_table_header *head,
 	if (ret == 0 && ctx.new_updated) {
 		kfree(*buf);
 		*buf = ctx.new_val;
-		*pcount = ctx.new_len;
+		if (!(*buf)[ctx.new_len])
+			*pcount = ctx.new_len - 1;
+		else
+			*pcount = ctx.new_len;
 	} else {
 		kfree(ctx.new_val);
 	}
-- 
2.43.0