Ilia Mirkin <imir...@alum.mit.edu> writes:
> On Mon, Mar 17, 2014 at 2:42 PM, Ilia Mirkin <imir...@alum.mit.edu> wrote:
>> There are a lot of different pci ids supported by nouveau, and more are
>> added all the time. The relevant distinguisher between drivers is the
>> chipset id.
>>
>> Signed-off-by: Ilia Mirkin <imir...@alum.mit.edu>
>> Cc: "10.1" <mesa-sta...@lists.freedesktop.org>
>> ---
>>
>> I haven't tested this on real hardware yet, but perhaps this is a little less
>> hacky than the former version and will actually work for more than NV05. When
>> I test it out tonight, I'll add a print in there to make sure that it's
>> hitting the expected paths.
>>
>> src/loader/Makefile.sources | 3 +-
>> src/loader/loader.c | 5 +-
>> src/loader/pci_id_driver_map.c | 68
>> ++++++++++++++++++++++
>> .../pci_ids => src/loader}/pci_id_driver_map.h | 7 ++-
>> 4 files changed, 80 insertions(+), 3 deletions(-)
>> create mode 100644 src/loader/pci_id_driver_map.c
>> rename {include/pci_ids => src/loader}/pci_id_driver_map.h (90%)
>>
>> diff --git a/src/loader/Makefile.sources b/src/loader/Makefile.sources
>> index 51a64ea..1a1345f 100644
>> --- a/src/loader/Makefile.sources
>> +++ b/src/loader/Makefile.sources
>> @@ -1,2 +1,3 @@
>> LOADER_C_FILES := \
>> - loader.c
>> \ No newline at end of file
>> + loader.c \
>> + pci_id_driver_map.c
>> diff --git a/src/loader/loader.c b/src/loader/loader.c
>> index 811f8a2..e343f4a 100644
>> --- a/src/loader/loader.c
>> +++ b/src/loader/loader.c
>> @@ -78,7 +78,7 @@
>> #endif
>>
>> #define __IS_LOADER
>> -#include "pci_ids/pci_id_driver_map.h"
>> +#include "pci_id_driver_map.h"
>>
>> static void default_logger(int level, const char *fmt, ...)
>> {
>> @@ -352,6 +352,9 @@ loader_get_driver_for_fd(int fd, unsigned driver_types)
>> if (!(driver_types & driver_map[i].driver_types))
>> continue;
>>
>> + if (driver_map[i].predicate && !driver_map[i].predicate(fd))
>> + continue;
>> +
>> if (driver_map[i].num_chips_ids == -1) {
>> driver = strdup(driver_map[i].driver);
>> goto out;
>> diff --git a/src/loader/pci_id_driver_map.c b/src/loader/pci_id_driver_map.c
>> new file mode 100644
>> index 0000000..e5ba9fe
>> --- /dev/null
>> +++ b/src/loader/pci_id_driver_map.c
>> @@ -0,0 +1,68 @@
>> +/*
>> + * Copyright 2014 Ilia Mirkin
>> + *
>> + * Permission is hereby granted, free of charge, to any person obtaining a
>> + * copy of this software and associated documentation files (the
>> "Software"),
>> + * to deal in the Software without restriction, including without limitation
>> + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
>> + * and/or sell copies of the Software, and to permit persons to whom the
>> + * Software is furnished to do so, subject to the following conditions:
>> + *
>> + * The above copyright notice and this permission notice (including the next
>> + * paragraph) shall be included in all copies or substantial portions of the
>> + * Software.
>> + *
>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
>> OR
>> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
>> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
>> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
>> OTHER
>> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
>> FROM,
>> + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
>> IN THE
>> + * SOFTWARE.
>> + */
>> +
>> +#include <string.h>
>> +
>> +#ifndef __NOT_HAVE_DRM_H
>> +
>> +#include <xf86drm.h>
>> +#include <libdrm/nouveau_drm.h>
>> +
>> +int is_nouveau(int fd);
>> +int is_nouveau_vieux(int fd);
>> +
>> +static int
>> +nouveau_chipset(int fd)
>> +{
>> + struct drm_nouveau_getparam gp;
>> + int chipset, ret;
>> +
>> + memset(&gp, 0, sizeof(gp));
>> + gp.param = NOUVEAU_GETPARAM_CHIPSET_ID;
>> + gp.value = &chipset;
>> + ret = drmCommandWriteRead(fd, DRM_NOUVEAU_GETPARAM, &gp, sizeof(gp));
>
> ... and this doesn't work because the GETPARAM ioctl is marked as
> DRM_UNLOCKED|DRM_AUTH|DRM_RENDER_ALLOW and we aren't authed at this
> point, nor are we considered a render client. I repeat my earlier
> comment: ARGH! [And of course fixing this in the kernel is too late,
> since this has to work for older kernels too.]
>
> Any other ideas? Having to fill out the pci tables would really suck.
> I'm back to the "hard-code prefer X driver name for nouveau" idea...I don't see a reason not to get the magic and do the auth before doing loader_get_driver_for_fd(). (though, how big would the pciids table for nouveau_vieux be to just hardcode it in pci_id_driver_map.h? There aren't any more new ones being made, right?)
pgpP5s4NX19Cw.pgp
Description: PGP signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
