On Tue, Apr 15, 2014 at 12:03 PM, Tim Kryger <tim.kryger@xxxxxxxxxx> wrote: > On Mon, Apr 14, 2014 at 6:42 PM, Andrew Bresticker > <abrestic@xxxxxxxxxxxx> wrote: >> If regulator_get_optional() returns EPROBE_DEFER, it indicates >> that the regulator may show up later (e.g. the DT property is >> present but the corresponding regulator may not have probed). >> Instead of continuing without the regulator, return EPROBE_DEFER >> from sdhci_add_host(). Also, fix regulator leaks in the error >> paths in sdhci_add_host(). >> >> Signed-off-by: Andrew Bresticker <abrestic@xxxxxxxxxxxx> >> --- >> drivers/mmc/host/sdhci.c | 19 ++++++++++++++++--- >> 1 file changed, 16 insertions(+), 3 deletions(-) > > This appears to be an improvement on Mike Looijmans patch: > > https://lkml.org/lkml/2014/4/7/34 > > The regulator_put() calls are appropriate but I wonder if we should > take this a step farther. Ulf is sure to point out that > mmc_regulator_get_supply() can be used to get regulators (though it > does stuff the pointers in host->mmc->supply.vmmc/vqmmc instead of > host->vmmc/vqmmc). However, that function doesn't put back the > reference to vmmc if the request for vqmmc returns EPROBE_DEFER. If > it did, it believe it could be used here to simplify the error > handling as all the regulator checks would happen up front. What do > you think? Seems reasonable. The put()s aren't necessary with mmc_regulator_get_supply() since it uses the devm_* API, however it doesn't return an error in case of EPROBE_DEFER when getting vqmmc, which is what we want in this case. > > -Tim -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
