From: Tom Rix <trix@xxxxxxxxxx> [ Upstream commit 472eb39103e885f302fd8fd6eff104fcf5503f1b ] clang static analysis flags this problem hpioctl.c:513:7: warning: Branch condition evaluates to a garbage value if (pci.ap_mem_base[idx]) { ^~~~~~~~~~~~~~~~~~~~ If there is a failure in the middle of the memory space loop, only some of the memory spaces need to be cleaned up. At the error handler, idx holds the number of successful memory spaces mapped. So rework the handler loop to use the old idx. There is a second problem, the memory space loop conditionally iomaps()/sets the mem_base so it is necessay to initize pci. Fixes: 719f82d3987a ("ALSA: Add support of AudioScience ASI boards") Signed-off-by: Tom Rix <trix@xxxxxxxxxx> Link: https://lore.kernel.org/r/20200913165230.17166-1-trix@xxxxxxxxxx Signed-off-by: Takashi Iwai <tiwai@xxxxxxx> Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx> --- sound/pci/asihpi/hpioctl.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/sound/pci/asihpi/hpioctl.c b/sound/pci/asihpi/hpioctl.c index 496dcde9715d6..9790f5108a166 100644 --- a/sound/pci/asihpi/hpioctl.c +++ b/sound/pci/asihpi/hpioctl.c @@ -343,7 +343,7 @@ int asihpi_adapter_probe(struct pci_dev *pci_dev, struct hpi_message hm; struct hpi_response hr; struct hpi_adapter adapter; - struct hpi_pci pci; + struct hpi_pci pci = { 0 }; memset(&adapter, 0, sizeof(adapter)); @@ -499,7 +499,7 @@ int asihpi_adapter_probe(struct pci_dev *pci_dev, return 0; err: - for (idx = 0; idx < HPI_MAX_ADAPTER_MEM_SPACES; idx++) { + while (--idx >= 0) { if (pci.ap_mem_base[idx]) { iounmap(pci.ap_mem_base[idx]); pci.ap_mem_base[idx] = NULL; -- 2.25.1