If tcmu_handle_completions() has to process a padding shorter than sizeof(struct tcmu_cmd_entry), the current call to tcmu_flush_dcache_range() with sizeof(struct tcmu_cmd_entry) as length param is wrong and causes crashes on e.g. ARM, because tcmu_flush_dcache_range() in this case calls flush_dcache_page(vmalloc_to_page(start)); with start being an invalid address above the end of the vmalloc'ed area. The fix is to use the maximum of remaining ring space and sizeof(struct tcmu_cmd_entry) as the length param. The patch was tested on kernel 4.19.118. See https://bugzilla.kernel.org/show_bug.cgi?id=208045#c10 Signed-off-by: Bodo Stroesser <bstroesser@xxxxxxxxxxxxxx> Tested-by: JiangYu <lnsyyj@xxxxxxxxxxx> --- drivers/target/target_core_user.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/drivers/target/target_core_user.c b/drivers/target/target_core_user.c index 3885ca532f8f..82e476d48194 100644 --- a/drivers/target/target_core_user.c +++ b/drivers/target/target_core_user.c @@ -1221,7 +1221,14 @@ static unsigned int tcmu_handle_completions(struct tcmu_dev *udev) struct tcmu_cmd_entry *entry = (void *) mb + CMDR_OFF + udev->cmdr_last_cleaned; - tcmu_flush_dcache_range(entry, sizeof(*entry)); + /* + * Flush max. up to end of cmd ring, since current entry might + * be a padding that is shorter than sizeof(*entry) + */ + size_t ring_left = head_to_end(udev->cmdr_last_cleaned, + udev->cmdr_size); + tcmu_flush_dcache_range(entry, ring_left < sizeof(*entry) ? + ring_left : sizeof(*entry)); if (tcmu_hdr_get_op(entry->hdr.len_op) == TCMU_OP_PAD) { UPDATE_HEAD(udev->cmdr_last_cleaned, -- 2.12.3
