This patch series fixes bugs reported by Andrei Warkentin involving signal handling in linux-user mode. The first is Andrei's first patch (though I have tweaked the commit message a little). The second patch is aimed at fixing the locking bug that Andrei noted, in a somewhat simpler way than his patches use.
The test cases Andrei provided both pass with these patches. (The tb_lock is a completely ill-thought-out idea anyway; maybe one day we'll actually sort out how multiple linux-user threads should interact with the TCG data structures. But for now this fixes an obvious bug.) Andrei Warkentin (1): page_check_range: don't bail out early after unprotecting page Peter Maydell (1): cpu-exec: Unlock tb_lock if we longjmp out of code generation cpu-exec.c | 7 +++++++ translate-all.c | 1 - 2 files changed, 7 insertions(+), 1 deletion(-) -- 1.9.0