Not much of a calltrace, it scrolled away because of hardlockup detector.
On the bright side, radeon KMS worked correctly and actually showed it.
$ addr2line -e vmlinux ffffffff812d207b
drivers/scsi/scsi_lib.c:1015
969 int scsi_init_io(struct scsi_cmnd *cmd, gfp_t gfp_mask)
970 {
971 int error = scsi_init_sgtable(cmd->request, &cmd->sdb, gfp_mask);
972 if (error)
973 goto err_exit;
974
975 if (blk_bidi_rq(cmd->request)) {
976 struct scsi_data_buffer *bidi_sdb = kmem_cache_zalloc(
977 scsi_sdb_cache, GFP_ATOMIC);
978 if (!bidi_sdb) {
979 error = BLKPREP_DEFER;
980 goto err_exit;
981 }
982
983 cmd->request->next_rq->special = bidi_sdb;
984 error = scsi_init_sgtable(cmd->request->next_rq, bidi_sdb,
985 GFP_ATOMIC);
986 if (error)
987 goto err_exit;
988 }
989
990 if (blk_integrity_rq(cmd->request)) {
991 struct scsi_data_buffer *prot_sdb = cmd->prot_sdb;
992 int ivecs, count;
993
994 BUG_ON(prot_sdb == NULL);
995 ivecs = blk_rq_count_integrity_sg(cmd->request);
996
997 if (scsi_alloc_sgtable(prot_sdb, ivecs, gfp_mask)) {
998 error = BLKPREP_DEFER;
999 goto err_exit;
1000 }
1001
1002 count = blk_rq_map_integrity_sg(cmd->request,
1003 prot_sdb->table.sgl);
1004 BUG_ON(unlikely(count > ivecs));
1005
1006 cmd->prot_sdb = prot_sdb;
1007 cmd->prot_sdb->table.nents = count;
1008 }
1009
1010 return BLKPREP_OK ;
1011
1012 err_exit:
1013 scsi_release_buffers(cmd);
1014 scsi_put_command(cmd);
1015 ===> cmd->request->special = NULL; <===
1016 return error;
1017 }
1018 EXPORT_SYMBOL(scsi_init_io);
--
