| Re: [hatari-devel] Preparing for next release 2.4 |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/hatari-devel Archives
]
- To: hatari-devel@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [hatari-devel] Preparing for next release 2.4
- From: Andreas Grabher <andreas_g86@xxxxxxxxxx>
- Date: Tue, 21 Jun 2022 21:57:54 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=icloud.com; s=1a1hai; t=1655841480; bh=k+hI8WFd8e4cDafaP7d9upfyAPq+7Az+yJioYPPKlB4=; h=From:Content-Type:Mime-Version:Date:Subject:Message-Id:To; b=a6W6Jz5bmOhMBWMxvO/LvFhpm1rIxbj6qMdNO47VHaFgM6Txq4upZJyp7CEBPgMgQ iPQwIL3jQ4g48sUqZ6JMwMYVjeEfZaOBfVYZfFwGpeajlzmr2Xe+SjTR6eFPSK8LwA IHN9gGVwMqDzEE6KkewPT5aXEbJa4AHiidKBud3URY4JUpXnkiFAXFMQw5HLdz0CQV +JzYn1pynL7sab0WsrNr0pEjKpPVEgtjXI34mP3gKevkb5o6601sb3/QfM9h57mdh5 ASItqeuU4QL42qrjOAPiMamE46mUI5GnVK8Z5ncVSfzaREQ+Vk7Y8RHJYXtCoCd/Ff dqJf3YSaVMr7g==
> Am 21.06.2022 um 08:58 schrieb Nicolas Pomarède <npomarede@xxxxxxxxxxxx>:
>
>
>> Le 20/06/2022 à 21:48, Laurent Sallafranque a écrit :
>> OK, I've done some tests with my previous DSP interrupts code :
>> - flaysid works well
>> - mars_dsp doesn't work either (it may probably be a falcon timing problem introduced with the latest core)
>> - xperience by abstract doesn't work either with the latest core (same as mars_dsp).
>> So, if we decide to revert back to my old interrupts code, flaysid works again, but we may have to retest many demos, games and utilities.
>> PS: I've tested a few demos, and I didn't notice any regression with my previous code and the latest hatari core.
>> It's as you want.
>
> Hi
>
> in that case if other agree, I would keep the change you added to the DSP, adding some notes in compatibility.html about the non working demos.
> Then we can fix the dsp later after 2.4 if released.
>
> Else, as you say it would mean we need to test again many demos/programs and maybe need to fix more regressions.
>
> Nicolas
>
Of course my code might contain mistakes, but I did follow the data sheet very carefully. For debugging I would start looking for stuck interrupts. The old code did automatically release interrupts. That was wrong for most types of interrupts. Maybe some interrupt is not released correctly on the Atari side of the DSP interface. That might have been masked by the old (wrong) behavior.
Best wishes,
Andreas