From 5a7b910f93f2712618807c9eaf97392dcaa07592 Mon Sep 17 00:00:00 2001
From: Matthew Heon <matthew.heon@pm.me>
Date: Wed, 1 Jul 2020 14:32:49 -0400
Subject: [PATCH] In rootless + cgroupsv1, resource limits are an error

However, in some cases (unset limits), we can completely remove
the limit and avoid errors. This works around a bug where the
Podman frontend is setting a Pids limit of 0 on some rootless
systems.

For now, this is only implemented for the PID limit. It can
easily be extended to other resource limits, but it is a fair bit
of code to do so, so I leave that exercise to someone else.

Fixes #6834

Signed-off-by: Matthew Heon <matthew.heon@pm.me>
---
 pkg/domain/infra/abi/system.go   | 27 ---------------------------
 pkg/specgen/generate/validate.go | 25 +++++++++++++++++++++----
 2 files changed, 21 insertions(+), 31 deletions(-)

diff --git a/pkg/domain/infra/abi/system.go b/pkg/domain/infra/abi/system.go
index 0511289ab0..b347fbafa3 100644
--- a/pkg/domain/infra/abi/system.go
+++ b/pkg/domain/infra/abi/system.go
@@ -8,7 +8,6 @@ import (
 	"os/exec"
 	"path/filepath"
 	"strconv"
-	"syscall"
 
 	"github.com/containers/common/pkg/config"
 	"github.com/containers/libpod/libpod/define"
@@ -146,32 +145,6 @@ func movePauseProcessToScope() error {
 	return utils.RunUnderSystemdScope(int(pid), "user.slice", "podman-pause.scope")
 }
 
-func setRLimits() error { // nolint:deadcode,unused
-	rlimits := new(syscall.Rlimit)
-	rlimits.Cur = 1048576
-	rlimits.Max = 1048576
-	if err := syscall.Setrlimit(syscall.RLIMIT_NOFILE, rlimits); err != nil {
-		if err := syscall.Getrlimit(syscall.RLIMIT_NOFILE, rlimits); err != nil {
-			return errors.Wrapf(err, "error getting rlimits")
-		}
-		rlimits.Cur = rlimits.Max
-		if err := syscall.Setrlimit(syscall.RLIMIT_NOFILE, rlimits); err != nil {
-			return errors.Wrapf(err, "error setting new rlimits")
-		}
-	}
-	return nil
-}
-
-func setUMask() { // nolint:deadcode,unused
-	// Be sure we can create directories with 0755 mode.
-	syscall.Umask(0022)
-}
-
-// checkInput can be used to verify any of the globalopt values
-func checkInput() error { // nolint:deadcode,unused
-	return nil
-}
-
 // SystemPrune removes unused data from the system. Pruning pods, containers, volumes and images.
 func (ic *ContainerEngine) SystemPrune(ctx context.Context, options entities.SystemPruneOptions) (*entities.SystemPruneReport, error) {
 	var systemPruneReport = new(entities.SystemPruneReport)
diff --git a/pkg/specgen/generate/validate.go b/pkg/specgen/generate/validate.go
index bb3ca99072..1a5e70ee0e 100644
--- a/pkg/specgen/generate/validate.go
+++ b/pkg/specgen/generate/validate.go
@@ -3,6 +3,7 @@ package generate
 import (
 	"github.com/containers/common/pkg/sysinfo"
 	"github.com/containers/libpod/pkg/cgroups"
+	"github.com/containers/libpod/pkg/rootless"
 	"github.com/containers/libpod/pkg/specgen"
 	"github.com/pkg/errors"
 )
@@ -23,6 +24,25 @@ func verifyContainerResources(s *specgen.SpecGenerator) ([]string, error) {
 		return warnings, nil
 	}
 
+	if rootless.IsRootless() {
+		// Interpreting 0 as unset - remove any unset limits
+		if s.ResourceLimits.Pids != nil && s.ResourceLimits.Pids.Limit == 0 {
+			s.ResourceLimits.Pids = nil
+		}
+
+		// We are guaranteed to be cgroups v1 + rootless. Resource
+		// limits are not acceptable. But if everything is unset, we can
+		// just remove the block entirely.
+		if s.ResourceLimits.Memory == nil && s.ResourceLimits.Memory == nil &&
+			s.ResourceLimits.CPU == nil && s.ResourceLimits.BlockIO == nil &&
+			s.ResourceLimits.HugepageLimits == nil && s.ResourceLimits.Network == nil &&
+			s.ResourceLimits.Rdma == nil && s.ResourceLimits.Pids == nil {
+			s.ResourceLimits = nil
+			return warnings, nil
+		}
+		return warnings, errors.New("resource limits cannot be set for rootless containers in cgroups v1 mode")
+	}
+
 	// Memory checks
 	if s.ResourceLimits.Memory != nil {
 		memory := s.ResourceLimits.Memory
@@ -70,10 +90,7 @@ func verifyContainerResources(s *specgen.SpecGenerator) ([]string, error) {
 
 	// Pids checks
 	if s.ResourceLimits.Pids != nil {
-		pids := s.ResourceLimits.Pids
-		// TODO: Should this be 0, or checking that ResourceLimits.Pids
-		// is set at all?
-		if pids.Limit > 0 && !sysInfo.PidsLimit {
+		if !sysInfo.PidsLimit {
 			warnings = append(warnings, "Your kernel does not support pids limit capabilities or the cgroup is not mounted. PIDs limit discarded.")
 			s.ResourceLimits.Pids = nil
 		}