2022-04-22 13:30:04

by Alexey Gladkov

Subject: [PATCH v1 1/4] ipc: Remove extra1 field abuse to pass ipc namespace

Eric Biederman pointed out that using .extra1 to pass ipc namespace
looks like an ugly hack and there is a better solution.

Link: https://lore.kernel.org/lkml/[email protected]/
Signed-off-by: Eric W. Biederman <[email protected]>
Signed-off-by: Alexey Gladkov <[email protected]>
---
ipc/ipc_sysctl.c | 26 ++++++++------------------
1 file changed, 8 insertions(+), 18 deletions(-)

diff --git a/ipc/ipc_sysctl.c b/ipc/ipc_sysctl.c
index 15210ac47e9e..eb7ba8e0a355 100644
--- a/ipc/ipc_sysctl.c
+++ b/ipc/ipc_sysctl.c
@@ -19,16 +19,11 @@
static int proc_ipc_dointvec_minmax_orphans(struct ctl_table *table, int write,
void *buffer, size_t *lenp, loff_t *ppos)
{
- struct ipc_namespace *ns = table->extra1;
- struct ctl_table ipc_table;
+ struct ipc_namespace *ns =
+ container_of(table->data, struct ipc_namespace, shm_rmid_forced);
int err;

- memcpy(&ipc_table, table, sizeof(ipc_table));
-
- ipc_table.extra1 = SYSCTL_ZERO;
- ipc_table.extra2 = SYSCTL_ONE;
-
- err = proc_dointvec_minmax(&ipc_table, write, buffer, lenp, ppos);
+ err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);

if (err < 0)
return err;
@@ -55,20 +50,15 @@ static int proc_ipc_auto_msgmni(struct ctl_table *table, int write,
static int proc_ipc_sem_dointvec(struct ctl_table *table, int write,
void *buffer, size_t *lenp, loff_t *ppos)
{
- struct ipc_namespace *ns = table->extra1;
- struct ctl_table ipc_table;
+ struct ipc_namespace *ns =
+ container_of(table->data, struct ipc_namespace, sem_ctls);
int ret, semmni;

- memcpy(&ipc_table, table, sizeof(ipc_table));
-
- ipc_table.extra1 = NULL;
- ipc_table.extra2 = NULL;
-
semmni = ns->sem_ctls[3];
ret = proc_dointvec(table, write, buffer, lenp, ppos);

if (!ret)
- ret = sem_check_semmni(current->nsproxy->ipc_ns);
+ ret = sem_check_semmni(ns);

/*
* Reset the semmni value if an error happens.
@@ -131,6 +121,8 @@ static struct ctl_table ipc_sysctls[] = {
.maxlen = sizeof(init_ipc_ns.shm_rmid_forced),
.mode = 0644,
.proc_handler = proc_ipc_dointvec_minmax_orphans,
+ .extra1 = SYSCTL_ZERO,
+ .extra2 = SYSCTL_ONE,
},
{
.procname = "msgmax",
@@ -237,7 +229,6 @@ bool setup_ipc_sysctls(struct ipc_namespace *ns)

} else if (tbl[i].data == &init_ipc_ns.shm_rmid_forced) {
tbl[i].data = &ns->shm_rmid_forced;
- tbl[i].extra1 = ns;

} else if (tbl[i].data == &init_ipc_ns.msg_ctlmax) {
tbl[i].data = &ns->msg_ctlmax;
@@ -250,7 +241,6 @@ bool setup_ipc_sysctls(struct ipc_namespace *ns)

} else if (tbl[i].data == &init_ipc_ns.sem_ctls) {
tbl[i].data = &ns->sem_ctls;
- tbl[i].extra1 = ns;
#ifdef CONFIG_CHECKPOINT_RESTORE
} else if (tbl[i].data == &init_ipc_ns.ids[IPC_SEM_IDS].next_id) {
tbl[i].data = &ns->ids[IPC_SEM_IDS].next_id;
--
2.33.2

2022-05-02 23:14:04

by Eric W. Biederman

Subject: Re: [PATCH v1 1/4] ipc: Remove extra1 field abuse to pass ipc namespace

Alexey Gladkov <[email protected]> writes:

> Eric Biederman pointed out that using .extra1 to pass ipc namespace
> looks like an ugly hack and there is a better solution.
>
> Link: https://lore.kernel.org/lkml/[email protected]/
> Signed-off-by: Eric W. Biederman <[email protected]>
> Signed-off-by: Alexey Gladkov <[email protected]>
> ---
> ipc/ipc_sysctl.c | 26 ++++++++------------------
> 1 file changed, 8 insertions(+), 18 deletions(-)
>
> diff --git a/ipc/ipc_sysctl.c b/ipc/ipc_sysctl.c
> index 15210ac47e9e..eb7ba8e0a355 100644
> @@ -55,20 +50,15 @@ static int proc_ipc_auto_msgmni(struct ctl_table *table, int write,
> static int proc_ipc_sem_dointvec(struct ctl_table *table, int write,
> void *buffer, size_t *lenp, loff_t *ppos)
> {
> - struct ipc_namespace *ns = table->extra1;
> - struct ctl_table ipc_table;
> + struct ipc_namespace *ns =
> + container_of(table->data, struct ipc_namespace, sem_ctls);
> int ret, semmni;
>
> - memcpy(&ipc_table, table, sizeof(ipc_table));
> -
> - ipc_table.extra1 = NULL;
> - ipc_table.extra2 = NULL;
> -
> semmni = ns->sem_ctls[3];
> ret = proc_dointvec(table, write, buffer, lenp, ppos);
>
> if (!ret)
> - ret = sem_check_semmni(current->nsproxy->ipc_ns);
> + ret = sem_check_semmni(ns);
^^^^^^^^^^^^^^^^^^^^^^^^^^^

Can you break this one line change into a separate patch?

It is a bug fix so that the entire function uses the same
ns value. I expect the change would read easier if the
change was separate.

>
> /*
> * Reset the semmni value if an error happens.

Eric