↓
 

Kernel – 5.0-rc1 Released – Breaks NVIDIA and VMware..

Adventures With Linux ™ Posted on by

Kernel 5.0-rc1 is out, and brief details are here:  http://lkml.iu.edu/hypermail/linux/kernel/1901.0/02536.html

Compiling NVIDIA 410.93 fails:

..........................
 CC [M] /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.o
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c: In function ‘os_get_current_time’:
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:434:5: error: implicit declaration of function ‘do_gettimeofday’; did you mean ‘efi_gettimeofday’? [-Werror=implicit-function-declaration]
do_gettimeofday(&tm);
^~~~~~~~~~~~~~~
efi_gettimeofday
cc1: some warnings being treated as errors
make[3]: *** [/usr/src/linux-5.0-rc1/scripts/Makefile.build:277: /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.o] Error 1
make[2]: *** [/usr/src/linux-5.0-rc1/Makefile:1553: _module_/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel] Error 2
make[2]: Leaving directory '/usr/src/linux-5.0-rc1'
make[1]: *** [Makefile:152: sub-make] Error 2
make[1]: Leaving directory '/usr/src/linux-5.0-rc1'
make: *** [Makefile:79: modules] Error 2

Change do_gettimeofday to efi_gettimeofday)
in NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c
(multiple occurrences..)
Then:

............................
CC [M] /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/nv-vtophys.o
CC [M] /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.o
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c: In function ‘os_get_current_time’:
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:434:22: error: passing argument 1 of ‘efi_gettimeofday’ from incompatible pointer type [-Werror=incompatible-pointer-types]
efi_gettimeofday(&tm);
^~~
In file included from /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/common/inc/nv-linux.h:196,
from /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:15:
/usr/src/linux-5.0-rc1/include/linux/efi.h:1025:50: note: expected ‘struct timespec64 *’ but argument is of type ‘struct timeval *’
extern void efi_gettimeofday (struct timespec64 *ts);
~~~~~~~~~~~~~~~~~~~^~
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c: In function ‘os_delay’:
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:549:22: error: passing argument 1 of ‘efi_gettimeofday’ from incompatible pointer type [-Werror=incompatible-pointer-types]
efi_gettimeofday(&tm_aux);
^~~~~~~
In file included from /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/common/inc/nv-linux.h:196,
from /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:15:
/usr/src/linux-5.0-rc1/include/linux/efi.h:1025:50: note: expected ‘struct timespec64 *’ but argument is of type ‘struct timeval *’
extern void efi_gettimeofday (struct timespec64 *ts);
~~~~~~~~~~~~~~~~~~~^~
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:583:30: error: passing argument 1 of ‘efi_gettimeofday’ from incompatible pointer type [-Werror=incompatible-pointer-types]
efi_gettimeofday(&tm_aux);
^~~~~~~
In file included from /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/common/inc/nv-linux.h:196,
from /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.c:15:
/usr/src/linux-5.0-rc1/include/linux/efi.h:1025:50: note: expected ‘struct timespec64 *’ but argument is of type ‘struct timeval *’
extern void efi_gettimeofday (struct timespec64 *ts);
~~~~~~~~~~~~~~~~~~~^~
cc1: some warnings being treated as errors
make[3]: *** [/usr/src/linux-5.0-rc1/scripts/Makefile.build:277: /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel/nvidia/os-interface.o] Error 1
make[2]: *** [/usr/src/linux-5.0-rc1/Makefile:1553: _module_/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-410.93/kernel] Error 2
make[2]: Leaving directory '/usr/src/linux-5.0-rc1'
make[1]: *** [Makefile:152: sub-make] Error 2
make[1]: Leaving directory '/usr/src/linux-5.0-rc1'
make: *** [Makefile:79: modules] Error 2

Similar error with 415.25..

..............................
CC [M] /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-415.25/kernel/nvidia/nv-vtophys.o
CC [M] /home/rgadsdon/kernel/NVIDIA-Linux-x86_64-415.25/kernel/nvidia/os-interface.o
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-415.25/kernel/nvidia/os-interface.c: In function ‘os_get_current_time’:
/home/rgadsdon/kernel/NVIDIA-Linux-x86_64-415.25/kernel/nvidia/os-interface.c:432:5: error: implicit declaration of function ‘do_gettimeofday’; did you mean ‘efi_gettimeofday’? [-Werror=implicit-function-declaration]
do_gettimeofday(&tm);
^~~~~~~~~~~~~~~
efi_gettimeofday
cc1: some warnings being treated as errors
.................................

With VMware 15.0.2:

.................................
Makefile:189: ================= WARNING ================
Makefile:190: 'SUBDIRS' will be removed after Linux 5.3
Makefile:191: Please use 'M=' or 'KBUILD_EXTMOD' instead
Makefile:192: ==========================================

Then, many warnings (more than with Kernel 4.20..) and then:

.................................
 CC [M] /tmp/modconfig-Jp8XVA/vmmon-only/common/cpuid.o
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c: In function ‘HostIF_EstimateLockedPageLimit’:
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c:1641:25: error: ‘totalram_pages’ redeclared as different kind of symbol
extern unsigned long totalram_pages;
^~~~~~~~~~~~~~
In file included from /tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c:42:
./include/linux/mm.h:52:29: note: previous definition of ‘totalram_pages’ was here
static inline unsigned long totalram_pages(void)
^~~~~~~~~~~~~~
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c: In function ‘HostIFReadUptimeWork’:
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c:1787:4: error: implicit declaration of function ‘do_gettimeofday’; did you mean ‘do_settimeofday64’? [-Werror=implicit-function-declaration]
do_gettimeofday(&tv);
^~~~~~~~~~~~~~~
do_settimeofday64
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c: In function ‘HostIF_MapUserMem’:
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c:3408:40: error: macro "access_ok" passed 3 arguments, but takes just 2
if (!access_ok(VERIFY_WRITE, p, size)) {
^
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c:3408:9: error: ‘access_ok’ undeclared (first use in this function)
if (!access_ok(VERIFY_WRITE, p, size)) {
^~~~~~~~~
/tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.c:3408:9: note: each undeclared identifier is reported only once for each function it appears in
cc1: some warnings being treated as errors
make[2]: *** [scripts/Makefile.build:277: /tmp/modconfig-Jp8XVA/vmmon-only/linux/hostif.o] Error 1
make[2]: *** Waiting for unfinished jobs....
In file included from /tmp/modconfig-Jp8XVA/vmmon-only/./common/vmx86.h:32,
from /tmp/modconfig-Jp8XVA/vmmon-only/linux/driver.h:32,
from /tmp/modconfig-Jp8XVA/vmmon-only/linux/driver.c:47:
/tmp/modconfig-Jp8XVA/vmmon-only/./include/x86msr.h:130: warning: "MSR_MISC_FEATURES_ENABLES" redefined
#define MSR_MISC_FEATURES_ENABLES 0x140
..................................
<  more warnings follow... >

And, for VMNET:

..................................
 CC [M] /tmp/modconfig-Jp8XVA/vmnet-only/netif.o
/tmp/modconfig-Jp8XVA/vmnet-only/userif.c: In function ‘VNetUserIfMapPtr’:
/tmp/modconfig-Jp8XVA/vmnet-only/userif.c:145:52: error: macro "access_ok" passed 3 arguments, but takes just 2
if (!access_ok(VERIFY_WRITE, (void *)uAddr, size) ||
^
/tmp/modconfig-Jp8XVA/vmnet-only/userif.c:145:9: error: ‘access_ok’ undeclared (first use in this function)
if (!access_ok(VERIFY_WRITE, (void *)uAddr, size) ||
^~~~~~~~~
/tmp/modconfig-Jp8XVA/vmnet-only/userif.c:145:9: note: each undeclared identifier is reported only once for each function it appears in
make[2]: *** [scripts/Makefile.build:277: /tmp/modconfig-Jp8XVA/vmnet-only/userif.o] Error 1
make[2]: *** Waiting for unfinished jobs....
............................
< more warnings follow... >

So ..  More investigation needed!

Robert Gadsdon.   January 6, 2019.

Comments

Kernel – 5.0-rc1 Released – Breaks NVIDIA and VMware.. — 5 Comments

  1. Patches for VMware 15.0.2, 14.1.5 and 12.5.9 are ready in https://github.com/mkubecek/vmware-host-modules (they have been, in fact, available in a testing branch (workstation-15.0.2-test) for few days already but I didn’t want to push them into the “official” branch as I didn’t know if the version is going to be 4.21-rc1 or 5.0-rc1).

    Reply ↓

  2. Change do_gettimeofday to efi_gettimeofday)
    in NVIDIA-Linux-x86_64-410.25…9x/kernel/nvidia/os-interface.c is not work..
    better is :
    +#include
    + void do_gettimeofday(struct timeval *tv);
    do_gettimeofday is remove for Kernel 5 timekeeper…h

    Reply ↓

  3. NVIDIA-Linux-x86_64-410….

    — Change do_gettimeofday to efi_gettimeofday)
    — in NVIDIA-Linux-x86_64-410.25…93/kernel/nvidia/os-interface.c not work..
    In kernel 5 remove do_gettimeofday from timekeep…h
    #include
    void do_gettimeofday(struct timeval *tv); ..work..
    -nv_drm_gem_object_unreference_unlocked….. to
    +nv_drm_gem_object_put_unlocked….

    Reply ↓

  4. I’m quite sure that I’ve probably over-complicated this patch but I threw it together quite haphazardly when trying out the new 5.0-rc1 back on release.
    And note that there could very well be some calculation errors to the time code, though I’ve not noticed anything odd yet.
    I made the patch for 415.25, but the patch also seems to work with 415.27 just fine.
    Here’s the patch: https://gitlab.com/snippets/1799795

    And please excuse me if there’s already a fix out there, I checked but didn’t find anything after some light searching.

    Reply ↓

    • Many thanks for this.. I can’t find any other fixes, so far.
      Tested OK on 415.27 and Kernel 5.0-rc3 (see latest article..)
      RG.

      Reply ↓

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

↑