On 11/5/2024 13:21, Borislav Petkov wrote:
On Wed, Oct 23, 2024 at 05:21:50PM +0000, Yazen Ghannam wrote:+static ssize_t smn_value_write(struct file *file, const char __user *userbuf, + size_t count, loff_t *ppos) +{ + u32 val; + int ret; + + ret = kstrtouint_from_user(userbuf, count, 0, &val); + if (ret) + return ret; + + add_taint(TAINT_USER, LOCKDEP_STILL_OK);That looks like a TAINT_CPU_OUT_OF_SPEC to me.
Makes sense.
+ ret = amd_smn_write(debug_node, debug_address, val); + if (ret) + return ret; + + return count; +} + +DEFINE_SHOW_STORE_ATTRIBUTE(smn_node); +DEFINE_SHOW_STORE_ATTRIBUTE(smn_address); +DEFINE_SHOW_STORE_ATTRIBUTE(smn_value); + static int amd_cache_roots(void) { u16 node, num_nodes = amd_num_nodes(); @@ -180,6 +257,12 @@ static int __init amd_smn_init(void) if (err) return err;+ debugfs_dir = debugfs_create_dir("amd_smn", arch_debugfs_dir);+ + debugfs_create_file("node", 0600, debugfs_dir, NULL, &smn_node_fops); + debugfs_create_file("address", 0600, debugfs_dir, NULL, &smn_address_fops); + debugfs_create_file("value", 0600, debugfs_dir, NULL, &smn_value_fops);Can we pls stick this behind a module param which is off by default? I don't want that crap exposed even in debugfs, by default. Thx.
Why the worry about it being in debugfs by default?
