Hi, Here is the current batch of fixes for MTTCG regressions for the next release candidate. I've dropped the speculative atomic cputlb patch and included 2 fixes to the console code. The first fixes the taking of the BQL (which I missed) and the second uses the exclusive mechanism directly in the main thread rather than queuing the work to run in the vCPU context. I kept them separate as they are logically discreet changes.
The user-exec and bsd-user fixes have been posted separately before in the various report threads. Finally I claim a checkpatch exception for Pranith's newline fix. It didn't seem worth splitting the string for the sake of a semicolon on column 81 ;-) Gerd, Do you want to take the console patches via your tree or have me include it in a pull request? Alex Bennée (5): user-exec: handle synchronous signals from QEMU gracefully bsd-user: align use of mmap_lock to that of linux-user ui/console: ensure do_safe_dpy_refresh holds BQL ui/console: use exclusive mechanism directly replay/replay.c: bump REPLAY_VERSION Pranith Kumar (1): tcg: Add a new line after incompatibility warning bsd-user/mmap.c | 13 +------------ bsd-user/qemu.h | 2 -- cpu-exec-common.c | 2 +- cpus.c | 2 +- replay/replay.c | 2 +- ui/console.c | 18 +++++++++++------- user-exec.c | 18 +++++++++++++++--- 7 files changed, 30 insertions(+), 27 deletions(-) -- 2.11.0
