Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commita74740f

Browse files
committed
Provide a way to control SysV shmem attach address in EXEC_BACKEND builds.
In standard non-Windows builds, there's no particular reason to care whataddress the kernel chooses to map the shared memory segment at. However,when building with EXEC_BACKEND, there's a risk that the chosen addresswon't be available in all child processes. Linux with ASLR enabled (whichit is by default) seems particularly at risk because it puts shmem segmentsinto the same area where it maps shared libraries. We can work aroundthat by specifying a mapping address that's outside the range whereshared libraries could get mapped. On x86_64 Linux, 0x7e0000000000seems to work well.This is only meant for testing/debugging purposes, so it doesn't seemnecessary to go as far as providing a GUC (or any user-visibledocumentation, though we might change that later). Instead, it's justcontrolled by setting an environment variable PG_SHMEM_ADDR to thedesired attach address.Back-patch to all supported branches, since the point here is toremove intermittent buildfarm failures on EXEC_BACKEND animals.Owners of affected animals will need to add a suitable setting ofPG_SHMEM_ADDR to their build_env configuration.Discussion:https://postgr.es/m/7036.1492231361@sss.pgh.pa.us
1 parentbfba563 commita74740f

File tree

1 file changed

+23
-2
lines changed

1 file changed

+23
-2
lines changed

‎src/backend/port/sysv_shmem.c

Lines changed: 23 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -102,8 +102,28 @@ static void *
102102
InternalIpcMemoryCreate(IpcMemoryKeymemKey,Sizesize)
103103
{
104104
IpcMemoryIdshmid;
105+
void*requestedAddress=NULL;
105106
void*memAddress;
106107

108+
/*
109+
* Normally we just pass requestedAddress = NULL to shmat(), allowing the
110+
* system to choose where the segment gets mapped. But in an EXEC_BACKEND
111+
* build, it's possible for whatever is chosen in the postmaster to not
112+
* work for backends, due to variations in address space layout. As a
113+
* rather klugy workaround, allow the user to specify the address to use
114+
* via setting the environment variable PG_SHMEM_ADDR. (If this were of
115+
* interest for anything except debugging, we'd probably create a cleaner
116+
* and better-documented way to set it, such as a GUC.)
117+
*/
118+
#ifdefEXEC_BACKEND
119+
{
120+
char*pg_shmem_addr=getenv("PG_SHMEM_ADDR");
121+
122+
if (pg_shmem_addr)
123+
requestedAddress= (void*)strtoul(pg_shmem_addr,NULL,0);
124+
}
125+
#endif
126+
107127
shmid=shmget(memKey,size,IPC_CREAT |IPC_EXCL |IPCProtection);
108128

109129
if (shmid<0)
@@ -203,10 +223,11 @@ InternalIpcMemoryCreate(IpcMemoryKey memKey, Size size)
203223
on_shmem_exit(IpcMemoryDelete,Int32GetDatum(shmid));
204224

205225
/* OK, should be able to attach to the segment */
206-
memAddress=shmat(shmid,NULL,PG_SHMAT_FLAGS);
226+
memAddress=shmat(shmid,requestedAddress,PG_SHMAT_FLAGS);
207227

208228
if (memAddress== (void*)-1)
209-
elog(FATAL,"shmat(id=%d) failed: %m",shmid);
229+
elog(FATAL,"shmat(id=%d, addr=%p, flags=0x%x) failed: %m",
230+
shmid,requestedAddress,PG_SHMAT_FLAGS);
210231

211232
/* Register on-exit routine to detach new segment before deleting */
212233
on_shmem_exit(IpcMemoryDetach,PointerGetDatum(memAddress));

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp