Closed robertlipe closed 1 year ago
I saw this as well and the issue itself should be fixed in the TFT_eSPI repository. Pixelix at the moment just uses the fork of this lib, provided by HajuSchulz. @HajuSchulz Can you take care?
I'll check the current state of TFT_eSPI regarding this over the weekend. In early S3 days only this somewhat hasty fork existed that I pinned.
FWIW, I work with a lot of chips (team RISC-V) and there are now LOTS of SoCs out there with gpios above their word size (32/64). I've fixed a lot of code in libs and OSes for dealing this this. You can usually find a formula that. Extract these bits into an array index to hit the right regurster and those bits into a bit number.
Ch32v307 is like a $2 chip with 80(!) Gpios. It's single core, but otherwise a superset of esp32s3 as it's a newer part.
It's just a reality that we embedded types need to be prepared to deal with these days.
On Thu, Jul 27, 2023, 1:37 PM Norbert Schulz @.***> wrote:
I'll check the current state of TFT_eSPI regarding this over the weekend. In early S3 days only this somewhat hasty fork existed that I pinned.
— Reply to this email directly, view it on GitHub https://github.com/BlueAndi/esp-rgb-led-matrix/issues/139#issuecomment-1654213697, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACCSD33C6TK2L2GKLCOWNRLXSKYPPANCNFSM6AAAAAA2X653QA . You are receiving this because you authored the thread.Message ID: @.***>
The GPIO access problem got fixed meanwhile in the official TFT_eSPI repository. I'll prepare a push request to replace the early pinned S3 support patched one with it.
Thanx, but it's probably a distraction for me personally. I don't NEED GPIOs > 32. I just noticed the warning and helped analyze it.
Let's focus on the partitioning/boot issues, I've admittedly starved it for attention and need to come back to it.
On Sun, Jul 30, 2023 at 6:42 AM Norbert Schulz @.***> wrote:
The GPIO access problem got fixed meanwhile in the official TFT_eSPI repository. I'll prepare a push request to replace the early pinned S3 support patched one with it.
— Reply to this email directly, view it on GitHub https://github.com/BlueAndi/esp-rgb-led-matrix/issues/139#issuecomment-1657117524, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACCSD3ZFTPPFISK5AI7FGZLXSZCDXANCNFSM6AAAAAA2X653QA . You are receiving this because you authored the thread.Message ID: @.***>
Thanks Robert and Norbert! :-)
This doesn't actually affect me. It's my civic duty to report it and help analyze it because the next person to be affected by it may not be as mellow about it. I'm just killing time waiting for stupid python in stupid platformio to burn two minutes to analyze my stupid dependencies the 11 billionth's time before it performs a two second stupid action. (Hmm. mybe I should work on that mellow... On with the show!)
Describe the bug The LilyGo configuration, like many that are possible on the ESP32S3 which has something like 45 GPIO registers has register numbers above 31 and thus, incapable of fitting into a processor word.
To Reproduce Steps to reproduce the behavior: ~/.platformio/penv/bin/pio run --upload-port $PORT --target upload -e lilygo-t-display-s3
Hold your eyes open and don't blink when the yellow text scrolls by.
Expected behavior Glorious, warning-free compilation.
Screenshots If applicable, add screenshots to help explain your problem.
Please complete the following information: ~/.platformio/penv/bin/pio --version PlatformIO Core, version 6.1.9
Additional context
Catch one warning in isolation and it's "obvious" what's going on.
.pio/libdeps/lilygo-t-display-s3/TFT_eSPI/Processors/TFT_eSPI_ESP32_S3.c: In member function 'uint8_t TFT_eSPI::readByte()':