Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] phy: inno-hdmi: 32-bit overflow in inno_hdmi_phy_pre_pll_recalc_rate()
--
commit ddb60d944975dab0de39c50bd23d32e884d239ae
Author: Michal Wilczynski [off-list ref]
phy: Add common Innosilicon HDMI PHY helpers
This commit introduces a shared helper library for the Innosilicon HDMI PHY
pre-PLL configuration, including table lookup, clock rate calculation helpers,
and register programming.
quoted hunk ↗ jump to hunk
diff --git a/drivers/phy/phy-inno-hdmi.c b/drivers/phy/phy-inno-hdmi.c
new file mode 100644
index 0000000000000..6900d4d83e464
--- /dev/null
+++ b/drivers/phy/phy-inno-hdmi.c
[ ... ]
quoted hunk ↗ jump to hunk
+unsigned long
+inno_hdmi_phy_pre_pll_recalc_rate(const struct inno_hdmi_phy_pre_pll *pll,
+ unsigned long parent_rate)
+{
+ unsigned int control, div_1, div_2, div_3, pclk_div_ab, pclk_div_cd;
+ unsigned int frac_div_7_0, frac_div_15_8, frac_div_23_16, div;
+ u8 nd, no_a, no_b, no_d;
+ unsigned long frac, rate;
+ u64 vco, vco_rate;
+ u16 nf;
+ int ret;
[ ... ]
quoted hunk ↗ jump to hunk
+ nd = div_1 & INNO_PRE_PLL_PRE_DIV_MASK;
+ nf = (div_2 & INNO_PRE_PLL_FB_DIV_11_8_MASK) << 8;
+ nf |= div_3;
+ vco = parent_rate * nf;
+
+ if (!(div_2 & INNO_PRE_PLL_FRAC_DIV_DISABLE)) {
+ ret = inno_read(pll, INNO_PRE_PLL_FRAC_DIV_7_0, &frac_div_7_0);
+ if (ret)
+ goto err;
+ ret = inno_read(pll, INNO_PRE_PLL_FRAC_DIV_15_8, &frac_div_15_8);
+ if (ret)
+ goto err;
+ ret = inno_read(pll, INNO_PRE_PLL_FRAC_DIV_23_16,
+ &frac_div_23_16);
+ if (ret)
+ goto err;
+
+ frac = frac_div_7_0 | (frac_div_15_8 << 8) |
+ (frac_div_23_16 << 16);
+ vco += DIV_ROUND_CLOSEST(parent_rate * frac,
+ BIT(INNO_FRAC_DIV_WIDTH));
+ }
[Severity: Medium]
Can the VCO calculations in inno_hdmi_phy_pre_pll_recalc_rate() in
drivers/phy/phy-inno-hdmi.c overflow on 32-bit architectures?
Both parent_rate and frac have type unsigned long, while nf is a u16 and vco
is a u64. On 32-bit targets where unsigned long is 32 bits wide,
parent_rate * nf is evaluated in 32-bit arithmetic before being widened to
u64. With a typical 24 MHz reference clock, any feedback divider nf >= 179
exceeds U32_MAX, whereas the 12-bit feedback divider reaches up to 757 in the
Innosilicon pre-PLL configuration tables.
Similarly, when the fractional divider is enabled, parent_rate * frac and
DIV_ROUND_CLOSEST() operate on 32-bit unsigned long values before adding to
vco. For a 24 MHz parent_rate and a 24-bit frac value, parent_rate * frac
requires up to 49 bits and overflows 32-bit unsigned long whenever frac >= 179.
Should parent_rate be cast to u64 in parent_rate * nf, and should the
fractional calculation use DIV_ROUND_CLOSEST_ULL((u64)parent_rate * frac,
BIT(INNO_FRAC_DIV_WIDTH)) to avoid 32-bit truncation?
--
Sashiko AI review · https://sashiko.dev/#/patchset/20261028-jh7110-inno-hdmi-phy-v6-0-f3379684d97e@samsung.com?part=2
--
linux-phy mailing list
linux-phy@lists.infradead.org
https://lists.infradead.org/mailman/listinfo/linux-phy