Re: [PATCH 3/5] drm/msm/dsi_pll_10nm: Fix bad VCO rate calculation and prescaler

From: AngeloGioacchino Del Regno
Date: Tue Feb 02 2021 - 13:49:55 EST


Il 02/02/21 19:45, Rob Clark ha scritto:
On Tue, Feb 2, 2021 at 6:32 AM AngeloGioacchino Del Regno
<angelogioacchino.delregno@xxxxxxxxxxxxxx> wrote:

Il 01/02/21 18:31, Rob Clark ha scritto:
On Mon, Feb 1, 2021 at 9:18 AM Rob Clark <robdclark@xxxxxxxxx> wrote:

On Mon, Feb 1, 2021 at 9:05 AM Rob Clark <robdclark@xxxxxxxxx> wrote:

On Mon, Feb 1, 2021 at 7:47 AM Rob Clark <robdclark@xxxxxxxxx> wrote:

On Mon, Feb 1, 2021 at 2:11 AM AngeloGioacchino Del Regno
<angelogioacchino.delregno@xxxxxxxxxxxxxx> wrote:

Il 31/01/21 20:50, Rob Clark ha scritto:
On Sat, Jan 9, 2021 at 5:51 AM AngeloGioacchino Del Regno
<angelogioacchino.delregno@xxxxxxxxxxxxxx> wrote:

The VCO rate was being miscalculated due to a big overlook during
the process of porting this driver from downstream to upstream:
here we are really recalculating the rate of the VCO by reading
the appropriate registers and returning a real frequency, while
downstream the driver was doing something entirely different.

In our case here, the recalculated rate was wrong, as it was then
given back to the set_rate function, which was erroneously doing
a division on the fractional value, based on the prescaler being
either enabled or disabled: this was actually producing a bug for
which the final VCO rate was being doubled, causing very obvious
issues when trying to drive a DSI panel because the actual divider
value was multiplied by two!

To make things work properly, remove the multiplication of the
reference clock by two from function dsi_pll_calc_dec_frac and
account for the prescaler enablement in the vco_recalc_rate (if
the prescaler is enabled, then the hardware will divide the rate
by two).

This will make the vco_recalc_rate function to pass the right
frequency to the (clock framework) set_rate function when called,
which will - in turn - program the right values in both the
DECIMAL_DIV_START_1 and the FRAC_DIV_START_{LOW/MID/HIGH}_1
registers, finally making the PLL to output the right clock.

Also, while at it, remove the prescaler TODO by also adding the
possibility of disabling the prescaler on the PLL (it is in the
PLL_ANALOG_CONTROLS_ONE register).
Of course, both prescaler-ON and OFF cases were tested.

This somehow breaks things on sc7180 (display gets stuck at first
frame of splash screen). (This is a setup w/ an ti-sn65dsi86 dsi->eDP
bridge)


First frame of the splash means that something is "a bit" wrong...
...like the DSI clock is a little off.

I don't have such hardware, otherwise I would've tried... but what you
describe is a bit strange.
Is there any other older qcom platform using this chip? Any other
non-qcom platform? Is the driver for the SN65DSI86 surely fine?
Anyway, as you know, I would never propose untested patches nor
partially working ones for any reason: I'm sorry that this happened.

I don't think there is anything publicly avail w/ sc7180 (yet.. but very soon)

The ti-sn65dsi86 bridge is used on a bunch of 845/850 devices (like
the snapdragon windows laptops).. and I think also the older 835
laptops.. ofc that doesn't mean that there isn't some bug, but I'd
guess maybe more likely that there is some small difference in DSI vs
older devices, or some cmd vs video mode difference.

Anyways, seems like the screen did eventually recover so that gives me
a bit of confidence to bisect this series, which I'll do a bit later
today.

fwiw, this series minus this patch, and everything looks ok.. let me
take a closer look at what changes with this patch

Btw, it looks like upstream, config->disable_prescaler is always
false.. I don't suppose you have anything WIP that changes this?


Regarding that one, I have tested the driver in both cases, with
and without prescaler enabled (both worked fine), then I have decided
to leave the prescaler option exactly as the previous default.

My plan about this was/still is:
1. Wait until this one gets merged (gives me time to also look
at the other billion patches that I've sent);
2. Add the prescaler option DT property and explain that it has
to be used only with "puny" displays (low resolution, low
clocks) as with "good ones", enabling the prescaler gives less
clock jitter (and some microamps more power consumption);
3. Add the Spread Spectrum Clock (SSC) functionality with related
DT properties.

Point 2 and 3 would go in the same series, unless someone does
N.2 before I do... and N.3 requires a bit of extensive testing,
which I have already partially started on the FxTec phone.

fwiw, this is the clk_summary diff with and without this patch:

------------------
270,282c270,282
< dsi0_pll_out_div_clk 1 1 0
887039941 0 0 50000 Y
< dsi0_pll_post_out_div_clk 0 0 0
221759985 0 0 50000 Y
< dsi0_pll_bit_clk 2 2 0
887039941 0 0 50000 Y
< dsi0_pclk_mux 1 1 0
887039941 0 0 50000 Y
< dsi0_phy_pll_out_dsiclk 1 1 0
147839991 0 0 50000 Y
< disp_cc_mdss_pclk0_clk_src 1 1 0
147839991 0 0 50000 Y
< disp_cc_mdss_pclk0_clk 1 1 0
147839991 0 0 50000 Y
< dsi0_pll_by_2_bit_clk 0 0 0
443519970 0 0 50000 Y
< dsi0_phy_pll_out_byteclk 1 1 0
110879992 0 0 50000 Y
< disp_cc_mdss_byte0_clk_src 2 2 0
110879992 0 0 50000 Y
< disp_cc_mdss_byte0_div_clk_src 1 1
0 55439996 0 0 50000 Y
< disp_cc_mdss_byte0_intf_clk 1 1
0 55439996 0 0 50000 Y
< disp_cc_mdss_byte0_clk 1 1 0
110879992 0 0 50000 Y
---
dsi0_pll_out_div_clk 1 1 0 887039978 0 0 50000 Y
dsi0_pll_post_out_div_clk 0 0 0 221759994 0 0 50000 Y
dsi0_pll_bit_clk 2 2 0 887039978 0 0 50000 Y
dsi0_pclk_mux 1 1 0 887039978 0 0 50000 Y
dsi0_phy_pll_out_dsiclk 1 1 0 147839997 0 0 50000 Y
disp_cc_mdss_pclk0_clk_src 1 1 0 147839997 0 0 50000 Y
disp_cc_mdss_pclk0_clk 1 1 0 147839997 0 0 50000 Y
dsi0_pll_by_2_bit_clk 0 0 0 443519989 0 0 50000 Y
dsi0_phy_pll_out_byteclk 1 1 0 110879997 0 0 50000 Y
disp_cc_mdss_byte0_clk_src 2 2 0 110879997 0 0 50000 Y
disp_cc_mdss_byte0_div_clk_src 1 1 0 55439999 0 0 50000 Y
disp_cc_mdss_byte0_intf_clk 1 1 0 55439999 0 0 50000 Y
disp_cc_mdss_byte0_clk 1 1 0 110879997 0 0 50000 Y
------------------



This is almost exactly what I saw on my devices as well, you get a
difference of "just some Hz" (which can be totally ignored), because
of how the calculation is done now.

Thing is, what you see as PIXEL and BYTE clocks *before* the change is
Linux thinking that your DSI is at that frequency, while the PLL will
output *half* the rate, which is exactly what the patch fixes.

"Fun" story is: the Xperia XZ1 (8998) and XZ (8996) have got the same
display... by lowering the DSI rate on the MSM8996 phone by half, I
get the same *identical* issues as the 8998 one without this patch.
The clocks all match between one and another, because.. it's.. the same
display, after all.

It is because of the aforementioned test that I have raised doubts about
the TI chip driver (or anything else really).. but then, anything is
possible.

It does look like, *so far* the TI bridge chip is only used on qc
platforms (according to grep'ing dts), so I suppose I can't rule out
bugs which cancel each other out. Although there are various other
bridges used (for ex, the sdm845 rb3 board has some dsi->hdmi bridge)


Argh...

I guess it would be useful if we could measure the clk somehow to
confirm that it is running at the rate we think it is..


I totally agree with you on this, I actually wanted to do it the proper
way, but then these clocks are really too high for my cheap oscilloscope
and I couldn't... :(

BR,
-R



In any case, just to be perfectly transparent, while being here waiting
for review, this patch series got tested on more smartphones, even ones
that I don't personally own, with different displays.

For your reference, here's a list (all MSM8998..):
- OnePlus 5 (1920x1080)
- F(x)Tec Pro 1 (2160x1080)
- Sony Xperia XZ1 Compact (1280x720)
- Sony Xperia XZ1 (1920x1080)
- Sony Xperia XZ Premium (3840x2160)


Yeah, no worries, I wasn't trying to imply that the patch was untested.


I know, of course!

Out of curiosity, are any of those video mode panels?

Yes and "also":
The FxTec Pro1 has a video mode panel, for which I'm trying to upstream
the driver...look here: https://lore.kernel.org/patchwork/patch/1365228/

The Xperia XZ Premium has a Sharp LS055D1SX04 panel under NT35950, which
can be configured as command or as video mode... I tried both modes, but
there is some issue with the DPU1/DSI drivers and *DUAL DSI*, as cmd
does work with some tearing, but video doesn't even start (downstream it
works).

So the only video mode panel that I could test is that BOE panel on the
FxTec phone (single dsi), which works just great.



Also, something (I assume DSI related) that I was testing on
msm-next-staging seems to have effected the colors on the panel (ie.
they are more muted).. which seems to persist across reboots (ie. when

So much "fun". This makes me think something about the PCC block doing
the wrong thing (getting misconfigured).

switching back to a good kernel), and interestingly if I reboot from a
good kernel I see part of the login prompt (or whatever was previously
on-screen) in the firmware ui screen !?! (so maybe somehow triggered
the display to think it is in PSR mode??)


From a fast read, the SN65DSI86 is on I2C.. giving it a wrong dsi clock
cannot produce (logically, at least) this, so I say that it is very
unlikely for this to be a consequence of the 10nm pll fixes...


Note that the bridge can also be programmed via dsi cmd mode packets,
which I believe is the case on the 835 laptops (or at least one of
them).. but all the things I have are using i2c as the control path.

...unless the bootloader is not configuring the DSI rates, but that's
also veeeeery unlikely (it always does, or it always does not).

I haven't looked at the bootloader display code, but booting back to
an old/good kernel didn't change anything.. even powering off didn't.
But the ghost image seemed to fade after some time, and by the next
morning it was fine. Which is strange. (But tbf, I'm more a gpu guy
who works on display only when necessary.. ie. a gpu without a display
isn't so much fun ;-))


OpenCL all the way! lol :D

On Qualcomm platforms, the first thing that I've ever done was to bring
up displays on 8974 Sony platforms... (we're talking about years ago).

I'm a lil more on the display side of things (but growing a beard while
waiting between a frame and another due to no GPU isn't so much fun
either!).

Not sure if that is caused by these patches, but if I can figure out
how to get the panel back to normal I can bisect. I think for now
I'll drop this series. Possibly it could be a
two-wrongs-makes-a-right situation that had things working before, but
I think someone from qcom who knows the DSI IP should take a look.


I would be happy if someone from Qualcomm takes a look: after all, there
is no documentation and they're the only ones that can verify this kind
of stuff. Please, Qualcomm.

Hopefully someone can take a look.

Besides that, if there's anything I can help with to solve this riddle,
I'm here for you.

Thanks, like I said I'll try applying the patches one by one and see
if I can narrow down what made the panel go funny, and we can go from
there

BR,
-R

Yours,
-- Angelo

BR,
-R


Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@xxxxxxxxxxxxxx>
---
drivers/gpu/drm/msm/dsi/pll/dsi_pll_10nm.c | 22 +++++++++-------------
1 file changed, 9 insertions(+), 13 deletions(-)

diff --git a/drivers/gpu/drm/msm/dsi/pll/dsi_pll_10nm.c b/drivers/gpu/drm/msm/dsi/pll/dsi_pll_10nm.c
index 8b66e852eb36..5be562dfbf06 100644
--- a/drivers/gpu/drm/msm/dsi/pll/dsi_pll_10nm.c
+++ b/drivers/gpu/drm/msm/dsi/pll/dsi_pll_10nm.c
@@ -165,11 +165,7 @@ static void dsi_pll_calc_dec_frac(struct dsi_pll_10nm *pll)

pll_freq = pll->vco_current_rate;

- if (config->disable_prescaler)
- divider = fref;
- else
- divider = fref * 2;
-
+ divider = fref;
multiplier = 1 << config->frac_bits;
dec_multiple = div_u64(pll_freq * multiplier, divider);
dec = div_u64_rem(dec_multiple, multiplier, &frac);
@@ -266,9 +262,11 @@ static void dsi_pll_ssc_commit(struct dsi_pll_10nm *pll)

static void dsi_pll_config_hzindep_reg(struct dsi_pll_10nm *pll)
{
+ struct dsi_pll_config *config = &pll->pll_configuration;
void __iomem *base = pll->mmio;
+ u32 val = config->disable_prescaler ? 0x0 : 0x80;

- pll_write(base + REG_DSI_10nm_PHY_PLL_ANALOG_CONTROLS_ONE, 0x80);
+ pll_write(base + REG_DSI_10nm_PHY_PLL_ANALOG_CONTROLS_ONE, val);
pll_write(base + REG_DSI_10nm_PHY_PLL_ANALOG_CONTROLS_TWO, 0x03);
pll_write(base + REG_DSI_10nm_PHY_PLL_ANALOG_CONTROLS_THREE, 0x00);
pll_write(base + REG_DSI_10nm_PHY_PLL_DSM_DIVIDER, 0x00);
@@ -499,17 +497,15 @@ static unsigned long dsi_pll_10nm_vco_recalc_rate(struct clk_hw *hw,
frac |= ((pll_read(base + REG_DSI_10nm_PHY_PLL_FRAC_DIV_START_HIGH_1) &
0x3) << 16);

- /*
- * TODO:
- * 1. Assumes prescaler is disabled
- */
multiplier = 1 << config->frac_bits;
- pll_freq = dec * (ref_clk * 2);
- tmp64 = (ref_clk * 2 * frac);
+ pll_freq = dec * ref_clk;
+ tmp64 = ref_clk * frac;
pll_freq += div_u64(tmp64, multiplier);
-
vco_rate = pll_freq;

+ if (config->disable_prescaler)
+ vco_rate = div_u64(vco_rate, 2);
+
DBG("DSI PLL%d returning vco rate = %lu, dec = %x, frac = %x",
pll_10nm->id, (unsigned long)vco_rate, dec, frac);

--
2.29.2