From 21cb1e63a594e36ff350fba41600190fb0a1f42b Mon Sep 17 00:00:00 2001 From: =?utf8?q?Marc-Andr=C3=A9=20Lureau?= Date: Mon, 6 Nov 2017 19:39:12 +0100 Subject: [PATCH] tpm-passthrough: workaround a possible race MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit The TPM backend processing thread has common shared variable race issues. (they should not be so easy to reach since guest interaction with the device is slow compared to host emulation) An obvious one is setting op_cancelled from device thread after calling write(cancel_fd). The backend thread may return before the device thread has set the variable. Instead set it before cancellation. Even if the write() failed, the end result is command get possibly cancelled (even if cancellation came from external sources it doesn't matter much). It's worth to consider removing the backend processing thread for now. Signed-off-by: Marc-André Lureau Reviewed-by: Stefan Berger Signed-off-by: Stefan Berger --- hw/tpm/tpm_passthrough.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/hw/tpm/tpm_passthrough.c b/hw/tpm/tpm_passthrough.c index 788be3847d..73554aa792 100644 --- a/hw/tpm/tpm_passthrough.c +++ b/hw/tpm/tpm_passthrough.c @@ -89,6 +89,7 @@ static int tpm_passthrough_unix_tx_bufs(TPMPassthruState *tpm_pt, bool is_selftest; const struct tpm_resp_hdr *hdr; + /* FIXME: protect shared variables or use other sync mechanism */ tpm_pt->tpm_op_canceled = false; tpm_pt->tpm_executing = true; *selftest_done = false; @@ -178,12 +179,11 @@ static void tpm_passthrough_cancel_cmd(TPMBackend *tb) */ if (tpm_pt->tpm_executing) { if (tpm_pt->cancel_fd >= 0) { + tpm_pt->tpm_op_canceled = true; n = write(tpm_pt->cancel_fd, "-", 1); if (n != 1) { error_report("Canceling TPM command failed: %s", strerror(errno)); - } else { - tpm_pt->tpm_op_canceled = true; } } else { error_report("Cannot cancel TPM command due to missing " -- 2.11.4.GIT