Add /proc/gpu#/resubmit_runlist API

Resubmits the runlist in an identical configuration. Causes the
runlist scheduler to:
1. Reload and cache timeslice and scale values from TSGs.
2. Restart scheduling from the head of the runlist [may cause a
   preempt to be scheduled for the currently-running task (?)].
3. Address (?) an errata on Turing where re-enabled channels are
   not always detected.

Above behavior tested on GV100 and partially tested on TU102.

1 files changed, 27 insertions, 0 deletions

diff --git a/runlist_procfs.c b/runlist_procfs.c
index 0087d90..f569c77 100644
--- a/runlist_procfs.c
+++ b/runlist_procfs.c
@@ -207,6 +207,33 @@ struct file_operations preempt_tsg_file_ops = {
         .llseek = default_llseek,
 };
 
+
+ssize_t resubmit_runlist_file_write(struct file *f, const char __user *buffer,
+                               size_t count, loff_t *off) {
+        uint32_t target_runlist;
+        // Passing 0 as the base to kstrtou32 indicates autodetect hex/octal/dec
+        int err = kstrtou32_from_user(buffer, count, 0, &target_runlist);
+        struct nvdebug_state *g = &g_nvdebug_state[file2gpuidx(f)];
+        if (err)
+                return err;
+
+        // Verify valid runlist (in terms of absolute maximums)
+        if (g->chip_id < NV_CHIP_ID_TURING && target_runlist > MAX_RUNLISTS_GF100)
+                return -ERANGE;
+        else if (g->chip_id < NV_CHIP_ID_AMPERE && target_runlist > MAX_RUNLISTS_TU102)
+                return -ERANGE;
+
+        if ((err = resubmit_runlist(g, target_runlist)))
+                return err;
+
+        return count;
+}
+
+struct file_operations resubmit_runlist_file_ops = {
+        .write = resubmit_runlist_file_write,
+        .llseek = default_llseek,
+};
+
 ssize_t disable_channel_file_write(struct file *f, const char __user *buffer,
                                    size_t count, loff_t *off) {
         uint32_t target_channel;