| Message ID | 20230310074940.3475703-1-alexander.stein@ew.tq-group.com (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Headers | show |
| Series | [1/1] clk: rs9: Fix suspend/resume | expand |
Quoting Alexander Stein (2023-03-09 23:49:40) > Disabling the cache in commit 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > without removing cache synchronization in resume path results in a > kernel panic as map->cache_ops is unset, due to REGCACHE_NONE. > Enable flat cache again to support resume again. num_reg_defaults_raw > is necessary to read the cache defaults from hardware. Some registers > are strapped in hardware and cannot be provided in software. > > Fixes: 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> > --- > There is also a series to not panic when using regcache_sync on > REGCACHE_NONE maps at [1]. And it was rejected? > > [1] https://lore.kernel.org/lkml/20230310073911.3470892-1-alexander.stein@ew.tq-group.com/T/#u >
Am Montag, 13. März 2023, 23:26:54 CET schrieb Stephen Boyd: > Quoting Alexander Stein (2023-03-09 23:49:40) > > > Disabling the cache in commit 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > > without removing cache synchronization in resume path results in a > > kernel panic as map->cache_ops is unset, due to REGCACHE_NONE. > > Enable flat cache again to support resume again. num_reg_defaults_raw > > is necessary to read the cache defaults from hardware. Some registers > > are strapped in hardware and cannot be provided in software. > > > > Fixes: 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> > > --- > > There is also a series to not panic when using regcache_sync on > > REGCACHE_NONE maps at [1]. > > And it was rejected? v2 has been accepted and is available at [1] in regmap's for-next branch. Best regards, Alexander [1] https://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap.git/ commit/?id=fd883d79e4dcd2417c2b80756f22a2ff03b0f6e0
Quoting Alexander Stein (2023-03-14 00:22:07) > Am Montag, 13. M�rz 2023, 23:26:54 CET schrieb Stephen Boyd: > > Quoting Alexander Stein (2023-03-09 23:49:40) > > > > > Disabling the cache in commit 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > > > without removing cache synchronization in resume path results in a > > > kernel panic as map->cache_ops is unset, due to REGCACHE_NONE. > > > Enable flat cache again to support resume again. num_reg_defaults_raw > > > is necessary to read the cache defaults from hardware. Some registers > > > are strapped in hardware and cannot be provided in software. > > > > > > Fixes: 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > > > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> > > > --- > > > There is also a series to not panic when using regcache_sync on > > > REGCACHE_NONE maps at [1]. > > > > And it was rejected? > > v2 has been accepted and is available at [1] in regmap's for-next branch. > So this still needs to be applied and backported to fix stable kernels?
Am Montag, 20. März 2023, 20:35:41 CET schrieb Stephen Boyd: > Quoting Alexander Stein (2023-03-14 00:22:07) > > > Am Montag, 13. M�rz 2023, 23:26:54 CET schrieb Stephen Boyd: > > > Quoting Alexander Stein (2023-03-09 23:49:40) > > > > > > > Disabling the cache in commit 2ff4ba9e3702 ("clk: rs9: Fix I2C > > > > accessors") > > > > without removing cache synchronization in resume path results in a > > > > kernel panic as map->cache_ops is unset, due to REGCACHE_NONE. > > > > Enable flat cache again to support resume again. num_reg_defaults_raw > > > > is necessary to read the cache defaults from hardware. Some registers > > > > are strapped in hardware and cannot be provided in software. > > > > > > > > Fixes: 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > > > > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> > > > > --- > > > > There is also a series to not panic when using regcache_sync on > > > > REGCACHE_NONE maps at [1]. > > > > > > And it was rejected? > > > > v2 has been accepted and is available at [1] in regmap's for-next branch. > > So this still needs to be applied and backported to fix stable kernels? IMHO yes, because without cache nothing will be stored/restored upon suspend/ resume. The other patch in regcache was just to ensure broken drivers (like this one in the current state) do not crash the kernel. Best regards, Alexander
Quoting Alexander Stein (2023-03-09 23:49:40) > Disabling the cache in commit 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > without removing cache synchronization in resume path results in a > kernel panic as map->cache_ops is unset, due to REGCACHE_NONE. > Enable flat cache again to support resume again. num_reg_defaults_raw > is necessary to read the cache defaults from hardware. Some registers > are strapped in hardware and cannot be provided in software. > > Fixes: 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> > --- Applied to clk-fixes
diff --git a/drivers/clk/clk-renesas-pcie.c b/drivers/clk/clk-renesas-pcie.c index 1e8b196e36f3..3377ee55c4f0 100644 --- a/drivers/clk/clk-renesas-pcie.c +++ b/drivers/clk/clk-renesas-pcie.c @@ -149,8 +149,9 @@ static int rs9_regmap_i2c_read(void *context, static const struct regmap_config rs9_regmap_config = { .reg_bits = 8, .val_bits = 8, - .cache_type = REGCACHE_NONE, + .cache_type = REGCACHE_FLAT, .max_register = RS9_REG_BCP, + .num_reg_defaults_raw = 0x8, .rd_table = &rs9_readable_table, .wr_table = &rs9_writeable_table, .reg_write = rs9_regmap_i2c_write,
Disabling the cache in commit 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") without removing cache synchronization in resume path results in a kernel panic as map->cache_ops is unset, due to REGCACHE_NONE. Enable flat cache again to support resume again. num_reg_defaults_raw is necessary to read the cache defaults from hardware. Some registers are strapped in hardware and cannot be provided in software. Fixes: 2ff4ba9e3702 ("clk: rs9: Fix I2C accessors") Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> --- There is also a series to not panic when using regcache_sync on REGCACHE_NONE maps at [1]. [1] https://lore.kernel.org/lkml/20230310073911.3470892-1-alexander.stein@ew.tq-group.com/T/#u drivers/clk/clk-renesas-pcie.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)