Re: [PATCH 1/2] MIPS: MSA: bugfix - disable MSA during thread switch correctly
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
- To: Ralf Baechle <ralf@xxxxxxxxxxxxxx>
- Subject: Re: [PATCH 1/2] MIPS: MSA: bugfix - disable MSA during thread switch correctly
- From: Leonid Yegoshin <Leonid.Yegoshin@xxxxxxxxxx>
- Date: Fri, 22 May 2015 12:06:08 -0700
- Cc: <linux-mips@xxxxxxxxxxxxxx>, <rusty@xxxxxxxxxxxxxxx>, <alexinbeijing@xxxxxxxxx>, <paul.burton@xxxxxxxxxx>, <david.daney@xxxxxxxxxx>, <alex@xxxxxxxxxxxxxxxx>, <linux-kernel@xxxxxxxxxxxxxxx>, <james.hogan@xxxxxxxxxx>, <markos.chandras@xxxxxxxxxx>, <macro@xxxxxxxxxxxxxx>, <eunb.song@xxxxxxxxxxx>, <manuel.lauss@xxxxxxxxx>, <andreas.herrmann@xxxxxxxxxxxxxxxxxx>
- In-reply-to: <555F776E.3070904@imgtec.com>
- User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.2.0
Ralf,
If there was TIF_USEDMSA in "prev" task then it means that all MSA HW is
in use.
And switch_to() checks this and transfers it to resume() to indicate
that MSA processing should be done.
Macro call "msa_save_all a0" right before disabling MSA in Config5
does a save of MSA registers. If it doesn't cause an exception then it
means that Config5 does exist and Config5.MIPS_CONF5_MSAEN does exist too.
- Leonid.
[Index of Archives]
[Linux MIPS Home]
[LKML Archive]
[Linux ARM Kernel]
[Linux ARM]
[Linux]
[Git]
[Yosemite News]
[Linux SCSI]
[Linux Hams]