Date: Tue, 28 Oct 2014 10:31:45 -0200 From: Luiz Otavio O Souza <lists.br@gmail.com> To: Dan Raymond <draymond@foxvalley.net> Cc: "freebsd-arm@freebsd.org" <freebsd-arm@freebsd.org> Subject: Re: new support for Raspberry Pi B+ Message-ID: <CAB=2f8x_m3Nc2TWvLoi-KftwfK42Ve6grQDiWkvjNqWyA4tnkQ@mail.gmail.com> In-Reply-To: <544D8D53.5060100@foxvalley.net> References: <544D8D53.5060100@foxvalley.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On 26 October 2014 22:09, Dan Raymond wrote: > One small correction to my previous test results: with the 16GB Transcend > card I *can* get it to boot reliably with the hw.bcm2835.sdhci.hs="0" hack. > > To summarize, I see no improvement regarding SD card failures between > r271779 and r273702. Nor do I see any improvement between the previous > commit to crochet and the current commit. I do see a minor regression: > without the sdhci.hs hack I used to get 50% boot failures on my 16GB > Transcend and now I get 100% boot failures. Dan, It's not a regression, it is a improvement, but as usual, it uncovered another bugs (a bug never come alone). Try 10 or 15 boots with and without the fix and check the differences on the SD card identification at dmesg, it should look like: mmcsd0: 2GB <SD SU02G 8.0 SN 07018D24 MFG 05/2013 by 3 SD> at mmc0 25.0MHz/4bit/65535-block Without the patch you'll see odd values for bus speed and width in some of the boots, they will coincide with the times that your card boots fine. What is happening here was that the controller cannot read the card identification data and it will switch to safe (and slow) defaults. With the patch the card is always identified with correct speed and bus speed, this leads to consistent failures as RPi cannot handle HS speed for some cards, whence the the use of hw.bcm2835.sdhci.hs="0" to make it work. So, to summarize: - The fix improve the card detection and identification, making it reliable. - The fix does not improve the situation of cards running into errors when working at HS. I'm still chasing other possible bugs in this driver, but no promises. Luiz
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAB=2f8x_m3Nc2TWvLoi-KftwfK42Ve6grQDiWkvjNqWyA4tnkQ>