Add seccomp sandbox for non-SFI NaCl

components/nacl/browser/nacl_process_host.cc

Index: components/nacl/browser/nacl_process_host.cc
diff --git a/components/nacl/browser/nacl_process_host.cc b/components/nacl/browser/nacl_process_host.cc
index 8857102c446e185db6638392d9d9289d2bd7f7a0..ae05426ef23ddc37e516b8eaf62311434d468414 100644
--- a/components/nacl/browser/nacl_process_host.cc
+++ b/components/nacl/browser/nacl_process_host.cc
@@ -582,6 +582,8 @@ bool NaClProcessHost::LaunchSelLdr() {
   CopyNaClCommandLineArguments(cmd_line.get());
 
   cmd_line->AppendSwitchASCII(switches::kProcessType,
+                              uses_nonsfi_mode_ ?

[Mark Seaborn, 2014/03/28 01:38:25]

Can you split out the parts that plumb through "uses_nonsfi_mode" to nacl_helper
into a separate change?

There's already a place where we're passing "uses_nonsfi_mode" to the
nacl_helper process.  nacl_listener.cc receives NaClProcessMsg_Start, which
contains this flag, but this isn't early enough for the seccomp-bpf sandboxing.

Rather than having two places where this is plumbed through, can you change
nacl_listener.cc to receive the flag via the new route you're adding? 
(Otherwise I'd worry that the two could get out of sync.)

[hamaji, 2014/03/28 12:06:10]

Thanks for the review! I've uploaded another CL for review.

https://codereview.chromium.org/216603002/

I'll update this CL based on your comments once it turned out the preparation
CLs seem to be OK.

+                              switches::kNaClNonSfiLoaderProcess :
                               switches::kNaClLoaderProcess);
   cmd_line->AppendSwitchASCII(switches::kProcessChannelID, channel_id);
   if (NaClBrowser::GetDelegate()->DialogsAreSuppressed())