Re: What's the AVR32 GCC in Microchip Studio and where's its source code?
On Sun, Oct 17, 2021 at 2:15 AM Gregory Nutt wrote: > > Those are pretty old. These are old too: > http://ww1.microchip.com/downloads/archive/avr32-gnu-toolchain-3.4.0.332-source.zip > > There are even old sources here: > https://www.microchip.com/en-us/development-tools-tools-and-software/avr-and-sam-downloads-archive > > There slightly newer gcc-avr32 RPMs available for various Linux > distributions (like *gcc*-*avr32*-linux-gnu-4.8.5-16.el7.1.x86_64.rpm for > CentOS 7). Maybe a source RPM could be modified? On FreeBSD we have these system packages available from the list below. All packages are built from sources in automated way (these are called Ports) and available as binary packaged with `pkg` utility, but still you can build everything from the sources from scratch. There are no binary blobs here, except firmware files for some dongles. Ity be safe to assume they are updated regularly. May add to common denominator of the Open-Source toolchain for AVR :-) % pkg search -x avr arduino-avrdude-6.3_3 Program for programming the on-chip memory of Atmel AVR Arduino CPUs avr-binutils-2.37_1,1 GNU binary tools avr-gcc-10.2.0 FSF GCC for Atmel AVR 8-bit RISC cross-development avr-gdb-7.3.1_6GNU GDB for the AVR target avr-libc-2.0.0_4,1 C and math library for the Atmel AVR controller family avra-1.4.2 Macro Assembler for Atmel AVR microcontrollers avrdude-6.3_3 Program for programming the on-chip memory of Atmel AVR CPUs avro-1.10.2Data serialization system avro-c-1.10.2 C library for Apache Avro avro-cpp-1.10.2C++ library for Apache Avro libpololu-avr-151002_1 Support libraries for Pololu robots simavr-1.7_1 Simulator for several Atmel AVR chips -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: What's the AVR32 GCC in Microchip Studio and where's its source code?
On Sun, Oct 17, 2021 at 12:15 PM Tomasz CEDRO wrote: > > On Sun, Oct 17, 2021 at 2:15 AM Gregory Nutt wrote: > > > > Those are pretty old. These are old too: > > http://ww1.microchip.com/downloads/archive/avr32-gnu-toolchain-3.4.0.332-source.zip > > > > There are even old sources here: > > https://www.microchip.com/en-us/development-tools-tools-and-software/avr-and-sam-downloads-archive > > > > There slightly newer gcc-avr32 RPMs available for various Linux > > distributions (like *gcc*-*avr32*-linux-gnu-4.8.5-16.el7.1.x86_64.rpm for > > CentOS 7). Maybe a source RPM could be modified? > > On FreeBSD we have these system packages available from the list > below. All packages are built from sources in automated way (these are > called Ports) and available as binary packaged with `pkg` utility, but > still you can build everything from the sources from scratch. There > are no binary blobs here, except firmware files for some dongles. > > Ity be safe to assume they are updated regularly. May add to common > denominator of the Open-Source toolchain for AVR :-) > > % pkg search -x avr > arduino-avrdude-6.3_3 Program for programming the on-chip > memory of Atmel AVR Arduino CPUs > avr-binutils-2.37_1,1 GNU binary tools > avr-gcc-10.2.0 FSF GCC for Atmel AVR 8-bit RISC > cross-development > avr-gdb-7.3.1_6GNU GDB for the AVR target > avr-libc-2.0.0_4,1 C and math library for the Atmel AVR > controller family > avra-1.4.2 Macro Assembler for Atmel AVR microcontrollers > avrdude-6.3_3 Program for programming the on-chip > memory of Atmel AVR CPUs > avro-1.10.2Data serialization system > avro-c-1.10.2 C library for Apache Avro > avro-cpp-1.10.2C++ library for Apache Avro > libpololu-avr-151002_1 Support libraries for Pololu robots > simavr-1.7_1 Simulator for several Atmel AVR chips Ah, you can check the Port details at www.freshports.org and sources at cgit.freebsd.org, where they are fetched from, how they are built, what patches are applied, etc. I used them loong time ago, also the MCS51 tools. When I am done with setting up NuttX and ESP32-C3 / RISC-V I will try the AVR, ARM, I have some bunch of the boards, and I have ATTiny10 purchased for some project but it has only 1K Flash so pure assembly needs to be used for this one :-) For instance here are port information: https://www.freshports.org/devel/avr-gcc/ Here is the GIT repository with sources: https://cgit.freebsd.org/ports/tree/ -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: ESP32-C3 RISC-V
On Sun, Oct 17, 2021 at 1:06 AM Abdelatif Guettouche wrote: > There isn't really a big difference between IDF's toolchain and the one > from Sifive, this is why we didn't add both. > Are you having issues running the one from SiFive? I know SiFive, I am following them directly at RISC-V lists, I am (individual) member of the organization :-) I would prefer to use system provided packages in the first place, then ESP-IDF second, then anything external that needs manual intervention. The goal is to create automation bootstrap.sh in the top level of my project source code so it sets up whole build environment and produce the firmware for clients with only one script. Using ESP-IDF toolchain gives me all tools necessary for that on various platforms and because these are the same tools different compiler problems are reduced to almost zero. I made this with MBED and Zephyr, it works quite well, I also want to have something like this with NuttX :-) > The CROSSDEV variable will just have the name of the toolchain. If we want > to add another toolchain we add another ifeq in > riscv/rv32im/Toolchains.defs. Thank you :-) I can see there are many variants of the RV. That would imply changing many files. Can we put that in some sort top level Toolchains.defs and then include that in machine variants configuration? Will try that and provide patches / pr :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: What's the AVR32 GCC in Microchip Studio and where's its source code?
I think that the naming is: - avr-gcc for the still common 8-bit AVR family - avr32-gcc for the nearly extinct 32-bit AVR32 family On Sun, Oct 17, 2021 at 4:16 AM Tomasz CEDRO wrote: > On Sun, Oct 17, 2021 at 2:15 AM Gregory Nutt wrote: > > > > Those are pretty old. These are old too: > > > http://ww1.microchip.com/downloads/archive/avr32-gnu-toolchain-3.4.0.332-source.zip > > > > There are even old sources here: > > > https://www.microchip.com/en-us/development-tools-tools-and-software/avr-and-sam-downloads-archive > > > > There slightly newer gcc-avr32 RPMs available for various Linux > > distributions (like *gcc*-*avr32*-linux-gnu-4.8.5-16.el7.1.x86_64.rpm for > > CentOS 7). Maybe a source RPM could be modified? > > On FreeBSD we have these system packages available from the list > below. All packages are built from sources in automated way (these are > called Ports) and available as binary packaged with `pkg` utility, but > still you can build everything from the sources from scratch. There > are no binary blobs here, except firmware files for some dongles. > > Ity be safe to assume they are updated regularly. May add to common > denominator of the Open-Source toolchain for AVR :-) > > % pkg search -x avr > arduino-avrdude-6.3_3 Program for programming the on-chip > memory of Atmel AVR Arduino CPUs > avr-binutils-2.37_1,1 GNU binary tools > avr-gcc-10.2.0 FSF GCC for Atmel AVR 8-bit RISC > cross-development > avr-gdb-7.3.1_6GNU GDB for the AVR target > avr-libc-2.0.0_4,1 C and math library for the Atmel AVR > controller family > avra-1.4.2 Macro Assembler for Atmel AVR > microcontrollers > avrdude-6.3_3 Program for programming the on-chip > memory of Atmel AVR CPUs > avro-1.10.2Data serialization system > avro-c-1.10.2 C library for Apache Avro > avro-cpp-1.10.2C++ library for Apache Avro > libpololu-avr-151002_1 Support libraries for Pololu robots > simavr-1.7_1 Simulator for several Atmel AVR chips > > -- > CeDeROM, SQ7MHZ, http://www.tomek.cedro.info >
Re: github nuttx/nuttx vs apache/incubator-nuttx
> Either way is fine, but we need action now. I have seen the same topic > happen on the email list many times, but nothing happens after that >discussion. I would be happy to delete the whole thing. Adbelatif has offered to delete the contents as well. But we would need consensus. A few people writing emails doesn't constitute consensus. That is why we have formal votes so that everyone's voice is heard and the direction is clear. With regard to the trademark usage, it would be good to get PPMC feedback on that. Perhaps they would have a clear position. I don't really know if there is an issue here or not... certainly there is less of an issue here than with NuttX YouTube, LinkedIN, Discord, and various NuttX named websites. The bitbucket nuttx repository would have the same trademark problem, if there is any. And there should be some position about where and how we provide OS tool support. If not github.com/nuttx, then where? Up until now no one has ever made any change to kconfig-frontends other than to fix compile failures due to toolchain differences. The only significant change has been to add Windows Native support and that is not in either the github or bitbucket repos.
Re: github nuttx/nuttx vs apache/incubator-nuttx
> I would be happy to delete the whole thing. Adbelatif has offered to > delete the contents as well. But we would need consensus. A few people > writing emails doesn't constitute consensus. That is why we have formal > votes so that everyone's voice is heard and the direction is clear. We can then create a VOTE thread to decide whether we want to delete it or not (can we consider this one as a DISCUSS thread?) > And there should be some position about where and how we provide OS tool > support. If not github.com/nuttx, then where? We've been told before that we can host well known projects even if they are in an incompatible license. We discussed moving them to Apache. I'm not sure if we want to revive that discussion as it always went unnoticed, but it would be ideal to have all the necessary repositories in one place. Otherwise, github.com/nuttx would be a good place, maybe we can also change the name to something similar to what Alin suggested above. On Sun, Oct 17, 2021 at 2:45 PM Gregory Nutt wrote: > > > Either way is fine, but we need action now. I have seen the same topic > > happen on the email list many times, but nothing happens after that > >discussion. > > I would be happy to delete the whole thing. Adbelatif has offered to > delete the contents as well. But we would need consensus. A few people > writing emails doesn't constitute consensus. That is why we have formal > votes so that everyone's voice is heard and the direction is clear. > > With regard to the trademark usage, it would be good to get PPMC feedback > on that. Perhaps they would have a clear position. I don't really know if > there is an issue here or not... certainly there is less of an issue here > than with NuttX YouTube, LinkedIN, Discord, and various NuttX named > websites. The bitbucket nuttx repository would have the same trademark > problem, if there is any. > > And there should be some position about where and how we provide OS tool > support. If not github.com/nuttx, then where? Up until now no one has ever > made any change to kconfig-frontends other than to fix compile failures due > to toolchain differences. The only significant change has been to add > Windows Native support and that is not in either the github or bitbucket > repos.
Re: github nuttx/nuttx vs apache/incubator-nuttx
On Sat, Oct 16, 2021 at 10:27 AM Tomasz CEDRO wrote: > Regarding kconfig-frontends I am in touch with Espressif and Debian > maintainers, also contacted initial author Yann MORIN. Maybe putting > this package into the Apache organization would be a good idea if > possible..? It occurred to me that if Yann MORIN doesn't want to maintain kconfig-frontends then perhaps we could ask for a one-time SGA and relicense; but then I think that kconfig was originally a Linux kernel thing that Yann MORIN and/or perhaps others managed to separate from the rest of the Linux build system and turn into a separate project. If so, then it probably falls under GPL and I don't know whether it is even feasible to relicense. Let's wait to hear from Yann MORIN, who will hopefully respond to your email, and see where to go from there. There is also another possibility of asking ASF for permission to host critical dependencies such as this and other tools under a separate repo. Normally this isn't done but exceptions have been made for projects with a long history and user base, e.g., Apache OpenOffice has its own website, openoffice.org, rather than being under the apache.org namespace. That was a specific exception for that project. Nathan
menuconfig
What I really like in NuttX that menuconfig has all available options in one place so newcomer can see how many possible choices are out there! :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: github nuttx/nuttx vs apache/incubator-nuttx
On Sun, Oct 17, 2021 at 9:10 AM Abdelatif Guettouche wrote: > > We've been told before that we can host well known projects even if > they are in an incompatible license. We discussed moving them to > Apache. > I'm not sure if we want to revive that discussion as it always went > unnoticed, but it would be ideal to have all the necessary > repositories in one place. I remember being told that as well. I think we will have to get permission to do it. Maybe it would be a good thing to have a [DISCUSS] and [VOTE] to determine whether the project wants to take that course, and if so, then we need to carefully present how/why we would like to host the tools. In addition to kconfig-frontends, there is also a customized version of GCC that is (IIRC) needed to make use of the FLAT (non-elf) binary loader (???). Nathan
Re: menuconfig
On Sun, Oct 17, 2021 at 4:33 PM Tomasz CEDRO wrote: > What I really like in NuttX that menuconfig has all available options > in one place so newcomer can see how many possible choices are out > there! :-) Just an idea: would it be possible to estimate final firmware size based on selected functionalities so user could see how much flash/ram selected functionality would take and what would be final firmware size? :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: github nuttx/nuttx vs apache/incubator-nuttx
The nxflat stuff in that repository is broken for modern gcc. I fixed it and have hosted it standalone here as well as added some basic CI. https://github.com/btashton/nxflat --Brennan On Sun, Oct 17, 2021, 7:38 AM Nathan Hartman wrote: > On Sun, Oct 17, 2021 at 9:10 AM Abdelatif Guettouche > wrote: > > > > We've been told before that we can host well known projects even if > > they are in an incompatible license. We discussed moving them to > > Apache. > > I'm not sure if we want to revive that discussion as it always went > > unnoticed, but it would be ideal to have all the necessary > > repositories in one place. > > > I remember being told that as well. > > I think we will have to get permission to do it. Maybe it would be a > good thing to have a [DISCUSS] and [VOTE] to determine whether the > project wants to take that course, and if so, then we need to > carefully present how/why we would like to host the tools. > > In addition to kconfig-frontends, there is also a customized version > of GCC that is (IIRC) needed to make use of the FLAT (non-elf) binary > loader (???). > > Nathan >
Re: What's the AVR32 GCC in Microchip Studio and where's its source code?
Yeah, Microchip never upstreamer the 32bit work and if you look at the source parts of it are very ugly and based on a quite old version of gcc. They do still host the source but it is behind a login wall. https://www.microchip.com/en-us/development-tools-tools-and-software/gcc-compilers-avr-and-arm --Brennan On Sun, Oct 17, 2021, 5:27 AM Gregory Nutt wrote: > I think that the naming is: > >- avr-gcc for the still common 8-bit AVR family >- avr32-gcc for the nearly extinct 32-bit AVR32 family > > > On Sun, Oct 17, 2021 at 4:16 AM Tomasz CEDRO wrote: > > > On Sun, Oct 17, 2021 at 2:15 AM Gregory Nutt wrote: > > > > > > Those are pretty old. These are old too: > > > > > > http://ww1.microchip.com/downloads/archive/avr32-gnu-toolchain-3.4.0.332-source.zip > > > > > > There are even old sources here: > > > > > > https://www.microchip.com/en-us/development-tools-tools-and-software/avr-and-sam-downloads-archive > > > > > > There slightly newer gcc-avr32 RPMs available for various Linux > > > distributions (like *gcc*-*avr32*-linux-gnu-4.8.5-16.el7.1.x86_64.rpm > for > > > CentOS 7). Maybe a source RPM could be modified? > > > > On FreeBSD we have these system packages available from the list > > below. All packages are built from sources in automated way (these are > > called Ports) and available as binary packaged with `pkg` utility, but > > still you can build everything from the sources from scratch. There > > are no binary blobs here, except firmware files for some dongles. > > > > Ity be safe to assume they are updated regularly. May add to common > > denominator of the Open-Source toolchain for AVR :-) > > > > % pkg search -x avr > > arduino-avrdude-6.3_3 Program for programming the on-chip > > memory of Atmel AVR Arduino CPUs > > avr-binutils-2.37_1,1 GNU binary tools > > avr-gcc-10.2.0 FSF GCC for Atmel AVR 8-bit RISC > > cross-development > > avr-gdb-7.3.1_6GNU GDB for the AVR target > > avr-libc-2.0.0_4,1 C and math library for the Atmel AVR > > controller family > > avra-1.4.2 Macro Assembler for Atmel AVR > > microcontrollers > > avrdude-6.3_3 Program for programming the on-chip > > memory of Atmel AVR CPUs > > avro-1.10.2Data serialization system > > avro-c-1.10.2 C library for Apache Avro > > avro-cpp-1.10.2C++ library for Apache Avro > > libpololu-avr-151002_1 Support libraries for Pololu robots > > simavr-1.7_1 Simulator for several Atmel AVR chips > > > > -- > > CeDeROM, SQ7MHZ, http://www.tomek.cedro.info > > >
Re: What's the AVR32 GCC in Microchip Studio and where's its source code?
Many thanks to all and for all the links - that's cleared the air a bit! embecosm: based on GCC-4.4.7 I'll add yet another AVR32 toolchain builder for those interested: https://gitlab.com/martinwguy/ct-ng-avr32 based on avr32-gnu-toolchain 4.3.2 (which seems to have disappeared from their releases!) forward-ported from gcc-4.4.3 binutils-1.18 newlib-1.16.0 to gcc-4.4.7 binutils.1.22 newlib-1.20.0 with some other patches to correct GCC pr's (from Debian and elsewhere) to fix incorrect code generation and for smaller code size, to fix dlmalloc in newlib which corrupts memory when using more than one region of RAM, and more hacks for smaller code size when using -Os It's based on crosstool-ng, so contains patches to mainline sources, not a modified gcc. Thanks for the pointer to RedHat, the only more recent version of gcc (4.8.5 over 4.4.7). If that will solve my company's problems I'll look at incorporating it, but I suspect it only produces better code, not language features. Blessings M
Re: github nuttx/nuttx vs apache/incubator-nuttx
I will go ahead and remove github.com/nuttx content. There is nother there of value now and everything is available elsewhere. So that is safe to do in any case.
Re: github nuttx/nuttx vs apache/incubator-nuttx
All of the junk repositories are gone (that is a good thing to do regardless of what we decide to do with the directory). The github.com/nuttx directory still exists. The following people have write access: - Alan Carvalho de Assis, acassis - Anthony Merlino,antmerlino - David Sidrane, davids5 - hartmannathan - liuguo09 - lAbdelatif Guettouche, Ouss4 - patacongo - Xiang Xiao, xiaoxiang781216 We can add more if we decide to use the directory. Or we can remove the directory if we decide we don't need it. While this does not resolve the important issue at hand, it does at least resolve newbies being confused as Tomasz was. On Sun, Oct 17, 2021 at 12:02 PM Gregory Nutt wrote: > I will go ahead and remove github.com/nuttx content. There is nother > there of value now and everything is available elsewhere. So that is safe > to do in any case. >
Re: github nuttx/nuttx vs apache/incubator-nuttx
TANK U SIR! =) On Sun, Oct 17, 2021 at 8:17 PM Gregory Nutt wrote: > > All of the junk repositories are gone (that is a good thing to do > regardless of what we decide to do with the directory). The > github.com/nuttx directory still exists. The following people have write > access: > >- Alan Carvalho de Assis, acassis >- Anthony Merlino,antmerlino >- David Sidrane, davids5 >- hartmannathan >- liuguo09 >- lAbdelatif Guettouche, Ouss4 >- patacongo >- Xiang Xiao, xiaoxiang781216 > > We can add more if we decide to use the directory. Or we can remove the > directory if we decide we don't need it. > > While this does not resolve the important issue at hand, it does at least > resolve newbies being confused as Tomasz was. > > On Sun, Oct 17, 2021 at 12:02 PM Gregory Nutt wrote: > > > I will go ahead and remove github.com/nuttx content. There is nother > > there of value now and everything is available elsewhere. So that is safe > > to do in any case. > > -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: ESP32-C3 RISC-V
On Sun, Oct 17, 2021 at 12:30 PM Tomasz CEDRO wrote:
> On Sun, Oct 17, 2021 at 1:06 AM Abdelatif Guettouche wrote:
> > The CROSSDEV variable will just have the name of the toolchain. If we want
> > to add another toolchain we add another ifeq in
> > riscv/rv32im/Toolchains.defs.
>
> Thank you :-) I can see there are many variants of the RV. That would
> imply changing many files. Can we put that in some sort top level
> Toolchains.defs and then include that in machine variants
> configuration?
>
> Will try that and provide patches / pr :-)
I have found 1.5 solution :-)
One quick solution is to simply pass `CROSSDEV` to make: `make
-DCROSSDEV=riscv32-esp-elf-`.
Another half of the solution is to put autodetection mechanism like
this in `riscv/rv32im/Toolchains.defs`:
RVTOOLCHAINS := \
riscv32-esp-elf-\
riscv32-unknown-elf-\
riscv64-none-elf-\
riscv64-unknown-elf-
RVTOOLCHAINS_AVAILABLE := $(foreach rvtoolchain, $(RVTOOLCHAINS),\
$(if $(shell which $(rvtoolchain)gcc), $(rvtoolchain) ) )
CROSSDEV ?= $(word 1,${RVTOOLCHAINS_AVAILABLE})
$(info "USING CROSSDEV: ${CROSSDEV}")
But this is not the acceptable solution as this Makefile is called
over and over during build, each time autodetecting available
toolchains slowing down the build process.
Where could we put that autodetection so its called only once?
`tools/configure.sh`?
Also it would be nice if that variable could be set / changed with
`make menuconfig`, so probably stored in `.config` ?
Work in progress :-)
--
CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
esp32c3-devkit:ble boot loop
Hello world :-) By selecting toolkit by hand (make CROSSDEV=riscv32-esp-elf-) I made it to build the esp32c3-devkit:ble example. Both on FreeBSD and Linux. I am using ESP-IDF toolchain. After flashing I can see on terminal emulator that device is locked in a boot loop :-( Is there any additional setup required? Can this additional setup be put into a board configuration so it works out of the box? The result is exactly the same both with build from FreeBSD and Linux. Any hints welcome :-) Tomek ps/2: I have created and send FreeBSD Port update (2.5.1 -> 3.1) to esptool system package that supports esp32-c3 / risc-v :-) https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=259235 -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
I have managed to build and flash esp32c3-devkit:ble on FreeBSD :-) It gets a boot loop problem but the build seem to have worked just the same as on Linux Debian :-) I have tried `./tools/configure.sh -B sim:nsh` but that needs porting `genromfs` to FreeBSD. work in progres.. -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
On Mon, Oct 18, 2021 at 1:27 AM Tomasz CEDRO wrote: > > I have managed to build and flash esp32c3-devkit:ble on FreeBSD :-) > > It gets a boot loop problem but the build seem to have worked just the > same as on Linux Debian :-) > > I have tried `./tools/configure.sh -B sim:nsh` but that needs porting > `genromfs` to FreeBSD. > > work in progres.. genromfs was easy peasy :-) I am using upstream version 0.5.7 from https://github.com/chexum/genromfs 0xCFMX4% ls -al nuttx -rwxr-xr-x 1 XXX XXX 1516992 Oct 18 01:41 nuttx 0xCFMX4% ./nuttx login: admin password: User Logged-in! NuttShell (NSH) NuttX-10.1.0 MOTD: username=admin password=Administrator nsh> uname -a NuttX 10.1.0 9cb35d79cc-dirty Oct 18 2021 01:40:26 sim sim nsh> ls /: bin/ dev/ etc/ proc/ tmp/ nsh> :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
By the way, did anyone since 1998 of genromfs existence notice this part of Makefile that may cripple your system files ?! install-man: # genromfs 0.5 installed the man page in this file, # remove it before someone notices :) if [ -f $(PREFIX)$(bindir)/man8 ]; then \ rm -f $(PREFIX)$(bindir)/man8; \ fi mkdir -p $(PREFIX)$(mandir)/man8 install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ Well I noticed that. It will not replace existing file not even delete the single file of interest it will delete whole man pages directory as part of install !!! Very funny :-( I had a similar situation with modern Ubuntu Linux adduser utility that removed my whole home directory without asking just because it existed :-( https://www.tomek.cedro.info/why-linux-sux/ I can only confirm never run Linux or GPL outside dedicated snapshotted virtual machine. Never attach real world data to it. Best not to use it at all. -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
On Sun, Oct 17, 2021 at 9:32 PM Tomasz CEDRO wrote: > By the way, did anyone since 1998 of genromfs existence notice this > part of Makefile that may cripple your system files ?! > > install-man: > # genromfs 0.5 installed the man page in this file, > # remove it before someone notices :) > if [ -f $(PREFIX)$(bindir)/man8 ]; then \ > rm -f $(PREFIX)$(bindir)/man8; \ > fi > mkdir -p $(PREFIX)$(mandir)/man8 > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > Haha oops! That's not good at all!! Thanks for noticing and reporting it here. Would you be inclined to open a pull request? I'd do it myself right away but am away from a real computer. Thanks, Nathan
Re: FreeBSD / BSD
On Mon, Oct 18, 2021 at 4:03 AM Nathan Hartman wrote: > > On Sun, Oct 17, 2021 at 9:32 PM Tomasz CEDRO wrote: > > > By the way, did anyone since 1998 of genromfs existence notice this > > part of Makefile that may cripple your system files ?! > > > > install-man: > > # genromfs 0.5 installed the man page in this file, > > # remove it before someone notices :) > > if [ -f $(PREFIX)$(bindir)/man8 ]; then \ > > rm -f $(PREFIX)$(bindir)/man8; \ > > fi > > mkdir -p $(PREFIX)$(mandir)/man8 > > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > > > > Haha oops! That's not good at all!! Thanks for noticing and reporting it > here. > > Would you be inclined to open a pull request? I'd do it myself right away > but am away from a real computer. > > Thanks, > Nathan Already did: https://github.com/chexum/genromfs/pull/3 :-) That showed up in 2002 with version 0.5 :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
By the way, did anyone since 1998 of genromfs existence notice this > part of Makefile that may cripple your system files ?! > > install-man: > # genromfs 0.5 installed the man page in this file, > # remove it before someone notices :) > if [ -f $(PREFIX)$(bindir)/man8 ]; then \ > rm -f $(PREFIX)$(bindir)/man8; \ > fi > mkdir -p $(PREFIX)$(mandir)/man8 > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > > Well I noticed that. It will not replace existing file not even delete > the single file of interest it will delete whole man pages directory > as part of install !!! Very funny :-( > Are you sure? It looks like it removes the $bindir/man8 which is probably /usr/local/bin/man8. Then it installs the manpages correctly at $mandir/man8 That is probably okay.
Re: FreeBSD / BSD
> Would you be inclined to open a pull request? I'd do it myself right away > but am away from a real computer. > We don't have genromfs in a repository. We do have a snapshot here: https://bitbucket.org/nuttx/tools/src/master/genromfs-0.5.2.tar.gz That is the version I recommend to people because that is the version of romfs that I developed the NuttX ROMFS file system against.
Re: FreeBSD / BSD
On Mon, Oct 18, 2021 at 4:11 AM Gregory Nutt wrote: > > By the way, did anyone since 1998 of genromfs existence notice this > > part of Makefile that may cripple your system files ?! > > > > install-man: > > # genromfs 0.5 installed the man page in this file, > > # remove it before someone notices :) > > if [ -f $(PREFIX)$(bindir)/man8 ]; then \ > > rm -f $(PREFIX)$(bindir)/man8; \ > > fi > > mkdir -p $(PREFIX)$(mandir)/man8 > > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > > > > Well I noticed that. It will not replace existing file not even delete > > the single file of interest it will delete whole man pages directory > > as part of install !!! Very funny :-( > > > > Are you sure? It looks like it removes the $bindir/man8 which is probably > /usr/local/bin/man8. Then it installs the manpages correctly at > $mandir/man8 Well also $(bindir) is messed with $(mandir). I just removed that part in the patch :-) > That is probably okay. Far from okay to rm -rf anything as root without user knowledge especially at /usr location. Why not rm -rf / right away? :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
genromfs 0.5.7 port has been submitted, with local patch until upstream gets updated :-) https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=259242 -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
> > By the way, did anyone since 1998 of genromfs existence notice this > > > part of Makefile that may cripple your system files ?! > > > > > > install-man: > > > # genromfs 0.5 installed the man page in this file, > > > # remove it before someone notices :) > > > if [ -f $(PREFIX)$(bindir)/man8 ]; then \ > > > rm -f $(PREFIX)$(bindir)/man8; \ > > > fi > > > mkdir -p $(PREFIX)$(mandir)/man8 > > > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > > > > > > Well I noticed that. It will not replace existing file not even delete > > > the single file of interest it will delete whole man pages directory > > > as part of install !!! Very funny :-( > > > > > > > Are you sure? It looks like it removes the $bindir/man8 which is > probably > > /usr/local/bin/man8. Then it installs the manpages correctly at > > $mandir/man8 > > Well also $(bindir) is messed with $(mandir). I just removed that part > in the patch :-) > > > That is probably okay. > > Far from okay to rm -rf anything as root without user knowledge > especially at /usr location. Why not rm -rf / right away? :-) > > True, but it is not doing rm -rf, it is doing rm -f to remove a regular > file named man8. >
Re: FreeBSD / BSD
On Mon, Oct 18, 2021 at 4:31 AM Gregory Nutt wrote: > > > > By the way, did anyone since 1998 of genromfs existence notice this > > > > part of Makefile that may cripple your system files ?! > > > > > > > > install-man: > > > > # genromfs 0.5 installed the man page in this file, > > > > # remove it before someone notices :) > > > > if [ -f $(PREFIX)$(bindir)/man8 ]; then \ > > > > rm -f $(PREFIX)$(bindir)/man8; \ > > > > fi > > > > mkdir -p $(PREFIX)$(mandir)/man8 > > > > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > > > > > > > > Well I noticed that. It will not replace existing file not even delete > > > > the single file of interest it will delete whole man pages directory > > > > as part of install !!! Very funny :-( > > > > > > > > > > Are you sure? It looks like it removes the $bindir/man8 which is > > probably > > > /usr/local/bin/man8. Then it installs the manpages correctly at > > > $mandir/man8 > > > > Well also $(bindir) is messed with $(mandir). I just removed that part > > in the patch :-) > > > > > That is probably okay. > > > > Far from okay to rm -rf anything as root without user knowledge > > especially at /usr location. Why not rm -rf / right away? :-) > > > > True, but it is not doing rm -rf, it is doing rm -f to remove a regular > > file named man8. True, sorry, not the correct location anyway, I have trauma and will never trust Linux again after it wanted to remove my real world home directory full of real world work :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: FreeBSD / BSD
Wow, that change has been in the Makefile for 20 years. Here is the original bugfix: https://github.com/chexum/genromfs/commit/6d1b1e46bb7dbc25ca521c5dfcdc37d17dca8615 The bug was: -mkdir -p $(PREFIX)$(mandir) +mkdir -p $(PREFIX)$(mandir)/man8 install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ Which installed the manpage in a regular file called man8 (the use of bindir instead of mandir was also a bug it should have done nothing). And this one makes the code look like it does today: https://github.com/chexum/genromfs/commit/427b121e23cd21d9b9ddb6caee3eb1064369358d Which looks wrong to me: Why does the change use bindir? Must be some other issue that I can't see???
Re: FreeBSD / BSD
On Mon, Oct 18, 2021 at 4:57 AM Gregory Nutt wrote: > > Wow, that change has been in the Makefile for 20 years. > > Here is the original bugfix: > https://github.com/chexum/genromfs/commit/6d1b1e46bb7dbc25ca521c5dfcdc37d17dca8615 > > The bug was: > > -mkdir -p $(PREFIX)$(mandir) > +mkdir -p $(PREFIX)$(mandir)/man8 > install -m 644 genromfs.8 $(PREFIX)$(mandir)/man8/ > > Which installed the manpage in a regular file called man8 (the use of > bindir instead of mandir was also a bug it should have done nothing). > > And this one makes the code look like it does today: > https://github.com/chexum/genromfs/commit/427b121e23cd21d9b9ddb6caee3eb1064369358d > > Which looks wrong to me: Why does the change use bindir? Must be some > other issue that I can't see??? Probably a "fix" for bad install by 0.5 :-) https://github.com/chexum/genromfs/blob/4fdb9d00df13dfe52c28e8af058d38ce8bea89a1/Makefile install-man: mkdir -p $(PREFIX)/$(bindir) install -m 644 genromfs.8 $(PREFIX)/$(bindir)/man8 I would keep that file instead removing anything from $envvar / $clivar / usr as root, this is part of base OS, messing with /usr is wrong, this place is sacred, /usr/local is for local packages :-) Also prefix and PREFIX can be provided as command line and environment parameter so its not a constant location, is user controlled, and may differ between installations :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info
Re: github nuttx/nuttx vs apache/incubator-nuttx
Thanks for cleaning it out On Sun, Oct 17, 2021, 20:17 Gregory Nutt wrote: > All of the junk repositories are gone (that is a good thing to do > regardless of what we decide to do with the directory). The > github.com/nuttx directory still exists. The following people have write > access: > >- Alan Carvalho de Assis, acassis >- Anthony Merlino,antmerlino >- David Sidrane, davids5 >- hartmannathan >- liuguo09 >- lAbdelatif Guettouche, Ouss4 >- patacongo >- Xiang Xiao, xiaoxiang781216 > > We can add more if we decide to use the directory. Or we can remove the > directory if we decide we don't need it. > > While this does not resolve the important issue at hand, it does at least > resolve newbies being confused as Tomasz was. > > On Sun, Oct 17, 2021 at 12:02 PM Gregory Nutt wrote: > > > I will go ahead and remove github.com/nuttx content. There is nother > > there of value now and everything is available elsewhere. So that is > safe > > to do in any case. > > >
Re: esp32c3-devkit:ble boot loop
Is this loop from the 1st stage bootloader? If that is the case, you only need to flash the 2nd stage bootloader. You can get prebuilt binaries for that on github.com/espressif/esp-nutt-bootloader and then from NuttX you can flash with make download ESPTOOL_BINDIR=... You can also build it yourself if you want. Please check the documentation, there should be direct links. (Using just a phone now sorry) On Mon, Oct 18, 2021, 01:22 Tomasz CEDRO wrote: > Hello world :-) > > By selecting toolkit by hand (make CROSSDEV=riscv32-esp-elf-) I made > it to build the esp32c3-devkit:ble example. Both on FreeBSD and Linux. > I am using ESP-IDF toolchain. > > After flashing I can see on terminal emulator that device is locked in > a boot loop :-( > > Is there any additional setup required? > > Can this additional setup be put into a board configuration so it > works out of the box? > > The result is exactly the same both with build from FreeBSD and Linux. > > Any hints welcome :-) > Tomek > > ps/2: I have created and send FreeBSD Port update (2.5.1 -> 3.1) to > esptool system package that supports esp32-c3 / risc-v :-) > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=259235 > > > > -- > CeDeROM, SQ7MHZ, http://www.tomek.cedro.info >
