Re: [Keepalived-devel] error building 2.0.7
Status: Beta
Brought to you by:
acassen
Menu
▾
▴
Re: [Keepalived-devel] error building 2.0.7
|
From: Paul G. <gi...@tc...> - 2018-11-20 16:43:12
|
Hi, just so you know, the naming conflict above (with uClibc) is present in
the newest keepalived (2.0.10).
If anybody else is having this issue here is the patch I use for my system:
diff -Naur keepalived-2.0.10/lib/parser.c
keepalived-2.0.10_patch/lib/parser.c
--- keepalived-2.0.10/lib/parser.c 2018-11-05 21:49:48.000000000 +0000
+++ keepalived-2.0.10_patch/lib/parser.c 2018-11-20
15:55:08.381534860 +0000
@@ -1915,20 +1915,20 @@
#endif
if (!max_time)
- max_time = TIMER_MAX;
+ max_time = TIMER_MAXIMUM;
errno = 0;
timer = strtoul(vector_slot(strvec, index), &endptr, 10);
- *res = (timer > TIMER_MAX ? TIMER_MAX : timer) * TIMER_HZ;
+ *res = (timer > TIMER_MAXIMUM ? TIMER_MAXIMUM : timer) * TIMER_HZ;
if (FMT_STR_VSLOT(strvec, index)[0] == '-')
report_config_error(CONFIG_INVALID_NUMBER, "%snegative
number '%s'", warn, FMT_STR_VSLOT(strvec, index));
else if (*endptr)
report_config_error(CONFIG_INVALID_NUMBER, "%sinvalid
number '%s'", warn, FMT_STR_VSLOT(strvec, index));
- else if (errno == ERANGE || timer > TIMER_MAX)
+ else if (errno == ERANGE || timer > TIMER_MAXIMUM)
report_config_error(CONFIG_INVALID_NUMBER, "%snumber '%s'
outside timer range", warn, FMT_STR_VSLOT(strvec, index));
else if (timer < min_time || timer > max_time)
- report_config_error(CONFIG_INVALID_NUMBER, "number '%s'
outside range [%ld, %ld]", FMT_STR_VSLOT(strvec, index), min_time, max_time
? max_time : TIMER_MAX);
+ report_config_error(CONFIG_INVALID_NUMBER, "number '%s'
outside range [%ld, %ld]", FMT_STR_VSLOT(strvec, index), min_time, max_time
? max_time : TIMER_MAXIMUM);
else
return true;
diff -Naur keepalived-2.0.10/lib/parser.h
keepalived-2.0.10_patch/lib/parser.h
--- keepalived-2.0.10/lib/parser.h 2018-11-08 19:33:31.000000000 +0000
+++ keepalived-2.0.10_patch/lib/parser.h 2018-11-20
15:55:08.381534860 +0000
@@ -38,7 +38,7 @@
#define MAXBUF 1024
/* Maximum time read_timer can return */
-#define TIMER_MAX (ULONG_MAX / TIMER_HZ)
+#define TIMER_MAXIMUM (ULONG_MAX / TIMER_HZ)
/* Configuration test errors. These should be in decreasing order of
severity */
typedef enum {
On Thu, 18 Oct 2018 at 11:23, Paul Gildea <gi...@tc...> wrote:
> Apologies for the slow reply, I patched "#include <stdbool.h>" into
> core/process.c and keepalived-2.0.7 built.
> I haven't tested it out yet but if there is a problem with that solution
> (never really used C) I can work around it as it's just some definitions.
> I can log bugs about these if necessary? Even though it's more of a uClibc
> problem.
>
> Thanks,
>
> --
> Paul
>
> On Tue, 16 Oct 2018 at 17:39, Quentin Armitage <qu...@ar...>
> wrote:
>
>> Paul,
>>
>> It looks like bool, true and false are undeclared/undefined. So far as I
>> can see uClibc doesn't have a stdbool.h header file, but I'm not sure why
>> the #include <stdbool.h> isn't failing.
>>
>> Quentin
>>
>> On Tue, 2018-10-16 at 15:25 +0100, Paul Gildea wrote:
>>
>> Thanks Quentin, that cleared up the issue in my mind for me very well.
>> Either solution worked fine and build is progressing further.
>> It looks like there are some more issues for me so I'll see how I get on
>> and let you know of what happens:
>>
>> make[3]: Entering directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7new/keepalived/core'
>> CC main.o
>> CC daemon.o
>> CC pidfile.o
>> CC layer4.o
>> CC smtp.o
>> CC global_data.o
>> CC global_parser.o
>> CC process.o
>> process.c:41: error: expected '=', ',', ';', 'asm' or '__attribute__'
>> before 'realtime_priority_set'
>> process.c:49: error: expected '=', ',', ';', 'asm' or '__attribute__'
>> before 'priority_set'
>> process.c: In function 'set_process_priority':
>> process.c:79: error: 'priority_set' undeclared (first use in this
>> function)
>> process.c:79: error: (Each undeclared identifier is reported only once
>> process.c:79: error: for each function it appears in.)
>> process.c:79: error: 'true' undeclared (first use in this function)
>> process.c: In function 'reset_process_priority':
>> process.c:91: error: 'priority_set' undeclared (first use in this
>> function)
>> process.c:91: error: 'false' undeclared (first use in this function)
>> process.c: In function 'reset_process_priorities':
>> process.c:141: error: 'realtime_priority_set' undeclared (first use in
>> this function)
>> process.c:150: error: 'false' undeclared (first use in this function)
>> process.c:164: error: 'priority_set' undeclared (first use in this
>> function)
>> make[3]: *** [process.o] Error 1
>> make[3]: *** Waiting for unfinished jobs....
>> make[3]: Leaving directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7new/keepalived/core'
>> make[2]: *** [all-recursive] Error 1
>> make[2]: Leaving directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7new/keepalived'
>> make[1]: *** [all-recursive] Error 1
>> make[1]: Leaving directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7new'
>> make: ***
>> [/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7new/.stamp_built]
>> Error 2
>>
>>
>> On Tue, 16 Oct 2018 at 14:26, Quentin Armitage <qu...@ar...>
>> wrote:
>>
>> Paul,
>>
>> This appears probably to be due to a conflict between uClibc using
>> TIMER_MAX and keepalived defining TIMER_MAX. glibc doesn't have a
>> definition of, or use, TIMER_MAX, at least not on my systems, and musl
>> doesn't appear to either.
>>
>> So the problem would appear to be namespace polution by uClibc. Looking
>> at the uClibc-ng source code, TIMER_MAX doesn't appear to be used other
>> than to define it if it is not defined, in common/bits/uClibc_local_lim.h,
>> so I suspect it is some historical code in uClibc that should be removed.
>>
>> It would be simple enough to rename the TIMER_MAX used in keepalived to
>> something else, and there are only 4 lines in keepalived where it is used
>> (all in read_timer in lib/parser.c).
>>
>> Could you just try moving the definition of TIMER_MAX from lib/parser.h
>> to lib/parser.c and see if that works. If not, could you try changing it to
>> TIMER_MAX_UL or something like that and seeing if it resolves your problem.
>>
>> Please let me know what you work out, and we can then see if we can push
>> a commit to work around the problem in uClibc.
>>
>> Quentin Armitage
>> On Tue, 2018-10-16 at 12:22 +0100, Paul Gildea wrote:
>>
>> Hi,
>>
>> I currently have 2.0.5 build and was trying to patch it with a fix I need
>> by patching this commit to it:
>>
>> https://github.com/acassen/keepalived/commit/37d6b3bc22d307d09ab8bd75111c3645c742c792
>> However the code change must be too great between 2.0.5 and master
>> because doing so causes build errors. To make the code more similar before
>> patching I decided to upgrade to the latest release, 2.0.7, however that is
>> not building with the below error, do you know what might have changed to
>> cause this or what is happening? I'm not sure from reading the changelog. I
>> have also tried to build 2.0.6 to narrow it down, which worked fine.
>>
>> /media/vmpart/vrrp/buildroot/output/host/usr/x86_64-unknown-linux-uclibc/sysroot/usr/include/bits/uClibc_local_lim.h:15:3:
>> error: #error local_lim.h was incorrectly updated, use the NPTL version
>> from glibc
>> make[3]: *** [parser.o] Error 1
>> make[3]: *** Waiting for unfinished jobs....
>> CC logger.o
>> make[3]: Leaving directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7/lib'
>> make[2]: *** [all] Error 2
>> make[2]: Leaving directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7/lib'
>> make[1]: *** [all-recursive] Error 1
>> make[1]: Leaving directory
>> `/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7'
>> make: ***
>> [/media/vmpart/vrrp/buildroot/output/build/keepalived-2.0.7/.stamp_built]
>> Error 2
>>
>>
>> Looking at the file with that error I see:
>>
>> #if !defined PTHREAD_KEYS_MAX || defined TIMER_MAX || !defined
>> SEM_VALUE_MAX
>> # error local_lim.h was incorrectly updated, use the NPTL version from
>> glibc
>> #endif
>>
>> Regards,
>>
>> --
>> Paul
>>
>> _______________________________________________
>> Keepalived-devel mailing lis...@li...://lists.sourceforge.net/lists/listinfo/keepalived-devel
>>
>>
|
