[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Moving avrdude to Git ?
From: |
Hannes Weisbach |
Subject: |
Re: Moving avrdude to Git ? |
Date: |
Fri, 3 Dec 2021 09:26:21 +0100 |
> Am 03.12.2021 um 02:33 schrieb Xiaofan Chen <xiaofanc@gmail.com>:
>
> On Fri, Dec 3, 2021 at 7:29 AM Hannes Weisbach <hannes_weisbach@gmx.net>
> wrote:
>>
>> Just a comment about pthread dependencies, I think I’ve made before. If
>> nothing changed,
>> only the ftdi_syncbb programmer used that and it was required only, because
>> buffer limits
>> of the FTDI chip were ignored. Given proper handling of buffer sizes (see
>> avrftdi) no
>> deadlocks can occur using libftdi/FTDI chips and hence the need for
>> threading goes away.
>> I faintly remember making that argument before and advocating for the
>> pthreads
>> requirement be dropped because it allows for easier windows builds. I’m not
>> sure,
>> if I ever even submitted a patch for that, probably not.
>>
>> I know that this is a tangent to the current discussion, but if people start
>> investing time
>> in improving windows builds I thought I better mention it.
>
> If only libftdi is the constraint, there is an alternative solution to
> replace libftdi
> with libusb.
>
> FYI -- openocd project has done quite some conversion to replace libftdi
> with libusb.
> Eg:
> https://github.com/openocd-org/openocd/blob/master/src/jtag/drivers/mpsse.c
> https://github.com/openocd-org/openocd/blob/master/src/jtag/drivers/ft232r.c
With libusb you have the same issue that you would not be able to send
arbitrary amounts of data to the FTDI chip. That is the real issue in
ftdi_syncbb. Using threads is just a kludge to work around that.
—-
Hannes