Re: [arm-gnu] STM32 silicon errata

[Charles Manning <manningc2@xxxxxxxxxxxxx>]

That's a very silly policy on the part of ST.

Regardless of what compiler you are using you could be writing broken 
assembly.

How is anyone supposed to design a product using the part if they don't tell 
people what is broken. Admitting it is broken, but not being specific, is the 
worst possble thing to do PR wise.

Note to self: don't design in any ST parts...



On Friday 03 October 2008 20:43:50 Lanchon wrote:
> Thank you very much, Nathan. I already asked and ST does not want to
> answer. In fact, that erratum was only published after fuss on ST's
> forum. Regrettably, while working with ST I have grown accustomed to
> this kind of secrecy.
>
> Regards,
> Lanchon
>
> Nathan Sidwell wrote:
> > Lanchon,
> >
> >> ST makes the STM32 microcontroller line based on the Cortex-M3 core.
> >> Today ST published a new silicon errata document for the line in
> >> which it states that code generated by "GNU rev 4.2.3 and later" is
> >> incompatible with all but the latest silicon revision of the chips.
> >>
> >> 1) What new behavior in 4.2.3 triggers this issue?
> >> 2) What exactly goes wrong when the issue is triggered?
> >> 3) Was there a workaround in place for the STM32 in pre-4.2.3 compilers?
> >>
> >> Thank you very much for your help.
> >
> > You should ask ST for that information, the errata document you cite
> > has insufficient information to determine. CodeSourcery has no
> > additional information about this.
> >
> > nathan