Re: [PATCH] soc/tegra: fuse: Fix bitwise vs. logical OR warning
From: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Date: 2021-10-21 21:55:40
Also in:
lkml, llvm
On Thu, Oct 21, 2021 at 02:45:00PM -0700, Nathan Chancellor wrote: [...]
quoted hunk ↗ jump to hunk
--- a/drivers/soc/tegra/fuse/speedo-tegra20.c +++ b/drivers/soc/tegra/fuse/speedo-tegra20.c@@ -69,7 +69,7 @@ void __init tegra20_init_speedo_data(struct tegra_sku_info *sku_info) val = 0; for (i = CPU_SPEEDO_MSBIT; i >= CPU_SPEEDO_LSBIT; i--) { - reg = tegra_fuse_read_spare(i) | + reg = tegra_fuse_read_spare(i) || tegra_fuse_read_spare(i + CPU_SPEEDO_REDUND_OFFS); val = (val << 1) | (reg & 0x1); }@@ -84,7 +84,7 @@ void __init tegra20_init_speedo_data(struct tegra_sku_info *sku_info) val = 0; for (i = SOC_SPEEDO_MSBIT; i >= SOC_SPEEDO_LSBIT; i--) { - reg = tegra_fuse_read_spare(i) | + reg = tegra_fuse_read_spare(i) || tegra_fuse_read_spare(i + SOC_SPEEDO_REDUND_OFFS); val = (val << 1) | (reg & 0x1); }
It does seem correct, but nevertheless the code looks suspicious. reg is already masked with 0x1 as far as I can tell, and there are other places which depend on this (like speedo-tegra210.c). Guessing from the use of tegra_fuse_read_spare() I would recommend changing its return type as it is returing a bit value, not necessarily semantically a boolean. Best Regards Michał Mirosław