Drop has_program_compatibility_assistant flag
This commit is contained in:
@ -405,39 +405,35 @@ child_info_spawn::worker (const char *prog_arg, const char *const *argv,
|
||||
|
||||
c_flags |= CREATE_SEPARATE_WOW_VDM | CREATE_UNICODE_ENVIRONMENT;
|
||||
|
||||
if (wincap.has_program_compatibility_assistant ())
|
||||
/* We're adding the CREATE_BREAKAWAY_FROM_JOB flag here to workaround
|
||||
issues with the "Program Compatibility Assistant (PCA) Service".
|
||||
For some reason, when starting long running sessions from mintty(*),
|
||||
the affected svchost.exe process takes more and more memory and at one
|
||||
point takes over the CPU. At this point the machine becomes
|
||||
unresponsive. The only way to get back to normal is to stop the
|
||||
entire mintty session, or to stop the PCA service. However, a process
|
||||
which is controlled by PCA is part of a compatibility job, which
|
||||
allows child processes to break away from the job. This helps to
|
||||
avoid this issue.
|
||||
|
||||
First we call IsProcessInJob. It fetches the information whether or
|
||||
not we're part of a job 20 times faster than QueryInformationJobObject.
|
||||
|
||||
(*) Note that this is not mintty's fault. It has just been observed
|
||||
with mintty in the first place. See the archives for more info:
|
||||
http://cygwin.com/ml/cygwin-developers/2012-02/msg00018.html */
|
||||
JOBOBJECT_BASIC_LIMIT_INFORMATION jobinfo;
|
||||
BOOL is_in_job;
|
||||
|
||||
if (IsProcessInJob (GetCurrentProcess (), NULL, &is_in_job)
|
||||
&& is_in_job
|
||||
&& QueryInformationJobObject (NULL, JobObjectBasicLimitInformation,
|
||||
&jobinfo, sizeof jobinfo, NULL)
|
||||
&& (jobinfo.LimitFlags & (JOB_OBJECT_LIMIT_BREAKAWAY_OK
|
||||
| JOB_OBJECT_LIMIT_SILENT_BREAKAWAY_OK)))
|
||||
{
|
||||
/* We're adding the CREATE_BREAKAWAY_FROM_JOB flag here to workaround
|
||||
issues with the "Program Compatibility Assistant (PCA) Service"
|
||||
starting with Windows Vista. For some reason, when starting long
|
||||
running sessions from mintty(*), the affected svchost.exe process
|
||||
takes more and more memory and at one point takes over the CPU. At
|
||||
this point the machine becomes unresponsive. The only way to get
|
||||
back to normal is to stop the entire mintty session, or to stop the
|
||||
PCA service. However, a process which is controlled by PCA is part
|
||||
of a compatibility job, which allows child processes to break away
|
||||
from the job. This helps to avoid this issue.
|
||||
|
||||
First we call IsProcessInJob. It fetches the information whether or
|
||||
not we're part of a job 20 times faster than QueryInformationJobObject.
|
||||
|
||||
(*) Note that this is not mintty's fault. It has just been observed
|
||||
with mintty in the first place. See the archives for more info:
|
||||
http://cygwin.com/ml/cygwin-developers/2012-02/msg00018.html */
|
||||
|
||||
JOBOBJECT_BASIC_LIMIT_INFORMATION jobinfo;
|
||||
BOOL is_in_job;
|
||||
|
||||
if (IsProcessInJob (GetCurrentProcess (), NULL, &is_in_job)
|
||||
&& is_in_job
|
||||
&& QueryInformationJobObject (NULL, JobObjectBasicLimitInformation,
|
||||
&jobinfo, sizeof jobinfo, NULL)
|
||||
&& (jobinfo.LimitFlags & (JOB_OBJECT_LIMIT_BREAKAWAY_OK
|
||||
| JOB_OBJECT_LIMIT_SILENT_BREAKAWAY_OK)))
|
||||
{
|
||||
debug_printf ("Add CREATE_BREAKAWAY_FROM_JOB");
|
||||
c_flags |= CREATE_BREAKAWAY_FROM_JOB;
|
||||
}
|
||||
debug_printf ("Add CREATE_BREAKAWAY_FROM_JOB");
|
||||
c_flags |= CREATE_BREAKAWAY_FROM_JOB;
|
||||
}
|
||||
|
||||
if (mode == _P_DETACH)
|
||||
|
Reference in New Issue
Block a user