<html><head></head><body>Sounds like progress! Nice one!<br>
<br>
I encountered several issues with this box but still believe Cell can do more.<br>
<br>
We're at an incredible architecture right here, what concerns me most is that we still today miss the availability of the mightyness of this machine.<br>
<br>
Gallium 0.2 is still working for example, what we need here is people to bring this stuff further.<br>
<br>
Heck, I have 7 SPEs on this machine, time to do something useful with these!<br>
<br>
220Gflops FTW! <br><br><div class="gmail_quote">Am 31. Dezember 2017 23:50:14 MEZ schrieb Nathan Whitehorn <nwhitehorn@freebsd.org>:<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<pre class="k9mail"><br /><br />On 12/12/17 15:05, Geoff Levand wrote:<br /><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #729fcf; padding-left: 1ex;"> Hi Nathan,<br /><br /> On 12/08/2017 01:25 PM, Nathan Whitehorn wrote:<br /><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #ad7fa8; padding-left: 1ex;"> I submitted patches to libfdt that resolve this particular ABI breakage yesterday. If the patch gets merged, newer kernels should become bootable again.<br /></blockquote> Here's the link:<br /><br /> <a href="https://github.com/dgibson/dtc/pull/12">https://github.com/dgibson/dtc/pull/12</a> (Add limited read-only support for older (V2 and V3) device tree to libfdt.)<br /><br /> I rebased your patch and have it in my ps3-queue<br /> branch, but my ps3 with ps3-petitboot-09.11.30 still<br /> would not boot with it. I didn't spend any time yet<br /> to look into why.<br /><br /> <a href="https://git.kernel.org/pub/scm/linux/kernel/git/geoff/ps3-linux.git/log/?h=ps3-queue">https://git.kernel.org/pub/scm/linux/kernel/git/geoff/ps3-linux.git/log/?h=ps3-queue</a><br /><br /> -Geoff</blockquote><br /><br />I found a bug (and updated the pull request), which might help.<br /><br />The other potential issue is that the fdt_get_property*() API in libfdt <br />is fundamentally incompatible with the way that V2 device trees align <br />data accesses since it returns a pointer to a packed structure with both <br />the property header and data. However, the data in V2 devtrees may be <br />offset by 4 bytes in a way that depends both on the offset within the <br />tree and the length of the property and so cannot be represented only as <br />the tail of a fixed structure. I don't know if Linux uses those <br />functions; I had to replace two uses in FreeBSD with the equivalent <br />fdt_getprop*() calls to get things to work.<br />-Nathan<br /></pre></blockquote></div></body></html>