target/microblaze: Collected fixes for env->iflags
There are several problems here that can result in soft lockup,
depending on exactly where an interrupt or exception is delivered:
Include BIMM_FLAG in IFLAGS_TB_MASK, since it needs to follow D_FLAG.
Ensure that iflags is 0 when entering an interrupt/exception handler.
Add mb_cpu_synchronize_from_tb to restore iflags from tb->flags.
The change to t_sync_flags is cosmetic, but makes the code clearer.
This fixes the reported regression in acceptance/replay_kernel.py.
Fixes: 683a247ed7a4 ("target/microblaze: Store "current" iflags in insn_start")
Tested-by: Thomas Huth <thuth@redhat.com>
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20200904190842.2282109-2-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
diff --git a/target/microblaze/translate.c b/target/microblaze/translate.c
index a377818..a8a3249 100644
--- a/target/microblaze/translate.c
+++ b/target/microblaze/translate.c
@@ -91,8 +91,8 @@
static void t_sync_flags(DisasContext *dc)
{
/* Synch the tb dependent flags between translator and runtime. */
- if ((dc->tb_flags ^ dc->base.tb->flags) & ~MSR_TB_MASK) {
- tcg_gen_movi_i32(cpu_iflags, dc->tb_flags & ~MSR_TB_MASK);
+ if ((dc->tb_flags ^ dc->base.tb->flags) & IFLAGS_TB_MASK) {
+ tcg_gen_movi_i32(cpu_iflags, dc->tb_flags & IFLAGS_TB_MASK);
}
}
