1 files changed, 21 insertions, 0 deletions

diff --git a/gitlab/issues/target_missing/host_missing/accel_missing/1557.toml b/gitlab/issues/target_missing/host_missing/accel_missing/1557.toml
new file mode 100644
index 00000000..87fbe1dc
--- /dev/null
+++ b/gitlab/issues/target_missing/host_missing/accel_missing/1557.toml
@@ -0,0 +1,21 @@
+id = 1557
+title = "qemu-binfmt-conf.sh handles errors inconsistently"
+state = "opened"
+created_at = "2023-03-25T15:33:30.574Z"
+closed_at = "n/a"
+labels = ["linux-user"]
+url = "https://gitlab.com/qemu-project/qemu/-/issues/1557"
+host-os = "n/a"
+host-arch = "n/a"
+qemu-version = "n/a"
+guest-os = "n/a"
+guest-arch = "n/a"
+description = """We are installing qemu via multiarch/qemu-user-static docker image. https://github.com/multiarch/qemu-user-static
+
+What we have noticed is that because qemu-binfmt-conf.sh does not use `set -e`, its behavior with regards to failures is inconsistent. In short, registering the same thing into binfmt twice is an error (you get EEXIST). However, the exit code of qemu-binfmt-conf.sh itself seems to depend only on whether the last interpreter succeeded, leading to confusing and inconsistent results."""
+reproduce = """1. Register only qemu-arm-static interpreter with binfmt.
+2. Run qemu-binfmt-conf.sh. Observe that the exit code is zero, and logs show the duplicate interpreter was rejected.
+3. Remove all qemu interpreters.
+3. Register only qemu-loongarch64-static interpreter (currently last in qemu_target_list) with binfmt.
+3. Run qemu-binfmt-conf.sh. Observe that the exit code is non-zero, and logs show the duplicate interpreter was rejected."""
+additional = "n/a"