On Mon, Feb 09, 2026 at 10:46:03PM +0100, Marek Vasut wrote: > On 2/9/26 9:51 PM, Tom Rini wrote: > > On Mon, Feb 09, 2026 at 09:35:54PM +0100, Marek Vasut wrote: > > > On 2/9/26 9:14 PM, Jonas Karlman wrote: > > > > clk_get_rate() in U-Boot is documented to return clock rate on success, > > > > 0 for invalid clock and -ve error code for other errors. This differ > > > > slightly from Linux where only >= 0 is returned from clk_get_rate(). > > > > > > > > Some clock drivers take advantage of this difference and may return -ve > > > > error code for clocks not fully supported in U-Boot. > > > > > > > > Use IS_ERR_VALUE() to check for an error code in addition to current > > > > invalid clock check to fix broken and unpredicted behavior when clock > > > > driver returns a -ve error code for the ref_clk. > > > > > > > > Signed-off-by: Jonas Karlman <[email protected]> > > > I wonder if it would be better to fix the API discrepancy . Could you try > > > that instead ? > > > > Which API discrepancy? clk_get_rate behaves the way it behaves > > intentionally and there's some fairly long and recent threads about that > > resulting from attepmts to change that. > Align clk_get_rate() between U-Boot and Linux kind of discrepancy .
That's not in the current list of options. Please look at the various threads in the past 6-8 months (and some reverts about series that tried this). That was my initial opinion too, FWIW. -- Tom
signature.asc
Description: PGP signature
