Merge pull request #123 from emkey1/dev

Merge 1.2.4 Build 429 changes
emkey1 · Oct 30, 2022 · e9490b4 · e9490b4
2 parents 99bc216 + 02b6bd0
commit e9490b4
Show file tree

Hide file tree

Showing 33 changed files with 474 additions and 240 deletions.
diff --git a/app/AppDelegate.m b/app/AppDelegate.m
@@ -45,6 +45,10 @@ @interface AppDelegate ()
 static void ios_handle_exit(struct task *task, int code) {
     // we are interested in init and in children of init
     // this is called with pids_lock as an implementation side effect, please do not cite as an example of good API design
+    if(task->pid > MAX_PID) {// Corruption
+        printk("ERROR: Insane PID in ios_handle_exit(%d)\n", task->pid);
+        return;
+    }
     if (task->parent != NULL && task->parent->parent != NULL)
         return;
     // pid should be saved now since task would be freed

diff --git a/app/UserPreferences.m b/app/UserPreferences.m
@@ -7,6 +7,13 @@
 
 #import "UserPreferences.h"
 #import "fs/proc/ish.h"
+#include "sync.h"
+#include "task.h"
+
+// Stuff to allow for cleaning up when doEnableExtraLocking is disabled.  -mke
+extern bool doEnableExtraLocking;
+extern lock_t pids_lock;
+extern struct list alive_pids_list;
 
 // IMPORTANT: If you add a constant here and expose it via UserPreferences,
 // consider if it also needs to be exposed as a friendly preference and included
@@ -438,6 +445,12 @@ - (void)setShouldEnableExtraLocking:(BOOL)dim {
 }
 
 - (BOOL)validateShouldEnableExtraLocking:(id *)value error:(NSError **)error {
+    // Should set task->critical_region.count to 0 for all active processes when this is set to false.  Otherwise stuff blows up.  -mke
+    if(doEnableExtraLocking == true) {  // This needs to be the opposite of what you would expect because of reasons.  -mke
+        complex_lockt(&pids_lock, 0, __FILE__, __LINE__);
+        dword_t res = zero_critical_regions_count();
+        unlock(&pids_lock);
+    }
     return [*value isKindOfClass:NSNumber.class];
 }
 

diff --git a/app/main.m b/app/main.m
@@ -8,9 +8,20 @@
 #import <UIKit/UIKit.h>
 #import "AppDelegate.h"
 extern void run_at_boot(void);
+#import <Foundation/Foundation.h>
+#import <Foundation/NSProcessInfo.h>
+
+void disable_app_nap(void)
+{
+   if ([[NSProcessInfo processInfo] respondsToSelector:@selector(beginActivityWithOptions:reason:)])
+   {
+      [[NSProcessInfo processInfo] beginActivityWithOptions:0x00FFFFFF reason:@"Not sleepy and don't want to nap"];
+   }
+}
 
 int main(int argc, char * argv[]) {
     @autoreleasepool {
+        disable_app_nap();  // No napping I say. -mke
         run_at_boot();
         return UIApplicationMain(argc, argv, nil, NSStringFromClass([AppDelegate class]));
     }

diff --git a/deps/aports/community/x86/index.txt b/deps/aports/community/x86/index.txt
@@ -1 +1 @@
-APKINDEX-v3.14-2022-10-01.tar.gz
+APKINDEX-v3.14-2022-10-28.tar.gz
diff --git a/deps/aports/main/x86/index.txt b/deps/aports/main/x86/index.txt
@@ -1 +1 @@
-APKINDEX-v3.14-2022-10-09.tar.gz
+APKINDEX-v3.14-2022-10-28.tar.gz
diff --git a/emu/cpuid.h b/emu/cpuid.h
@@ -2,6 +2,7 @@
 #define CPUID_H
 
 #include "misc.h"
+extern bool isGlibC;
 
 static inline void do_cpuid(dword_t *eax, dword_t *ebx, dword_t *ecx, dword_t *edx) {
     dword_t leaf = *eax;
@@ -17,11 +18,20 @@ static inline void do_cpuid(dword_t *eax, dword_t *ebx, dword_t *ecx, dword_t *e
             *eax = 0x0; // say nothing about cpu model number
             *ebx = 0x0; // processor number 0, flushes 0 bytes on clflush
             *ecx = 0; // we support none of the features in ecx
-            *edx = (1 << 0) // fpu
+            if(isGlibC) {
+                *edx = (1 << 0) // fpu
                 | (1 << 15) // cmov
                 | (1 << 23) // mmx
+                | (1 << 25) // sse // musl handles sse being enabled badly.  -mke
                 | (1 << 26) // sse2
                 ;
+            } else {
+                *edx = (1 << 0) // fpu
+                | (1 << 15) // cmov
+                | (1 << 23) // mmx
+                | (1 << 26) // sse2
+                ;
+            }
             break;
     }
 }

diff --git a/emu/decode.h b/emu/decode.h
@@ -269,15 +269,18 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
                            READMODRM; VMOV(xmm_modrm_val, xmm_modrm_reg,128); break;
                 case 0x11: TRACEI("movupd xmm, xmm:modrm");
                            READMODRM; VMOV(xmm_modrm_reg, xmm_modrm_val,128); break;
-                case 0x12: TRACEI("movlpd xmm, xmm:modrm");
-                           READMODRM; V_OP(movl_pd, xmm_modrm_val, xmm_modrm_reg,128); break;
-                case 0x13: TRACEI("movlpd xmm:modrm, reg");
-                           READMODRM; V_OP(movl_mem_pd, xmm_modrm_val, modrm_reg,128); break;
+                case 0x12: TRACEI("movlpd xmm, modrm");
+                           READMODRM; V_OP(movl_p, modrm_val, xmm_modrm_reg,64); break;
+                case 0x13: TRACEI("movlpd modrm, xmm");
+                           READMODRM; V_OP(movl_pm, xmm_modrm_reg, modrm_val,64); break;
                 case 0x14: TRACEI("unpcklpd xmm, xmm:modrm");
                            READMODRM; V_OP(unpackl_pd, xmm_modrm_val, xmm_modrm_reg,128); break;
                 case 0x15: TRACEI("unpckhpd xmm, xmm:modrm");
                            READMODRM; V_OP(unpackh_pd, xmm_modrm_val, xmm_modrm_reg,128); break;
-
+                case 0x16: TRACEI("movhpd xmm, modrm");
+                           READMODRM; V_OP(movh_p, modrm_val, xmm_modrm_reg,64); break;
+                case 0x17: TRACEI("movhpd modrm, xmm");
+                           READMODRM; V_OP(movh_pm, xmm_modrm_reg, modrm_val,64); break;
                 case 0x2e: TRACEI("ucomisd xmm, xmm:modrm");
                            READMODRM; V_OP(single_ucomi, xmm_modrm_val, xmm_modrm_reg,64); break;
                 case 0x2f: TRACEI("comisd xmm, xmm:modrm");
@@ -427,8 +430,8 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
                            READMODRM; V_OP(mulu, xmm_modrm_val, xmm_modrm_reg, 128); break;
                 case 0xe6: TRACEI("cvttpd2dq xmm:modrm, xmm");
                            READMODRM; V_OP(cvttpd2dq, xmm_modrm_val, xmm_modrm_reg,64); break;
-		case 0xe7: TRACEI("movntdq xmm:modrm, xmm");
-                           READMODRM_MEM; VMOV(xmm_modrm_reg, xmm_modrm_val,128); break;
+                case 0xe7: TRACEI("movntdq xmm, xmm:modrm");
+                           READMODRM; VMOV(xmm_modrm_reg, xmm_modrm_val,128); break;
                 case 0xe8: TRACEI("psubsb xmm:modrm, xmm");
                            READMODRM; V_OP(subss_b, xmm_modrm_val, xmm_modrm_reg,128); break;
                 case 0xe9: TRACEI("psubsw xmm:modrm, xmm");
@@ -472,14 +475,18 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
                            READMODRM; VMOV(xmm_modrm_val, xmm_modrm_reg,128); break;
                 case 0x11: TRACEI("movups xmm, xmm:modrm");
                            READMODRM; VMOV(xmm_modrm_reg, xmm_modrm_val,128); break;
-
+                case 0x12: TRACEI("movlps xmm, modrm");
+                           READMODRM; V_OP(movl_p, modrm_val, xmm_modrm_reg,64); break;
+                case 0x13: TRACEI("movlps modrm, xmm");
+                           READMODRM; V_OP(movl_pm, xmm_modrm_reg, modrm_val,64); break;
                 case 0x14: TRACEI("unpcklps xmm, xmm:modrm");
                            READMODRM; V_OP(unpackl_ps, xmm_modrm_val, xmm_modrm_reg,128); break;
                 case 0x15: TRACEI("unpckhps xmm, xmm:modrm");
                            READMODRM; V_OP(unpackh_ps, xmm_modrm_val, xmm_modrm_reg,128); break;
-                case 0x16: TRACEI("movlhps xmm, xmm:modrm");
-                           READMODRM; V_OP(movlh_ps, xmm_modrm_val, xmm_modrm_reg,128); break;
-
+                case 0x16: TRACEI("movhps xmm, modrm");
+                           READMODRM; V_OP(movh_p, modrm_val, xmm_modrm_reg,64); break;
+                case 0x17: TRACEI("movhps modrm, xmm");
+                           READMODRM; V_OP(movh_pm, xmm_modrm_reg, modrm_val,64); break;
                 case 0x2e: TRACEI("ucomiss xmm, xmm:modrm");
                            READMODRM; V_OP(single_ucomi, xmm_modrm_val, xmm_modrm_reg,32); break;
                 case 0x2f: TRACEI("comiss xmm, xmm:modrm");
@@ -537,9 +544,8 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
 
                 case 0xe5: TRACEI("pmulhw mm:modrm, mm");
                            READMODRM; V_OP(mulu, mm_modrm_val, mm_modrm_reg,64); break;
-                case 0xe7: TRACEI("movntq mm:modrm, mm");
+                case 0xe7: TRACEI("movntq mm, mm:modrm");
                            READMODRM_MEM; VMOV(mm_modrm_reg, mm_modrm_val,64); break;
-
                 case 0xef: TRACEI("pxor mm:modrm, mm");
                            READMODRM; V_OP(xor, mm_modrm_val, mm_modrm_reg,64); break;
 

diff --git a/emu/memory.c b/emu/memory.c
@@ -44,10 +44,6 @@ void mem_init(struct mem *mem) {
 }
 
 void mem_destroy(struct mem *mem) {
-    //int elock_fail = 0;
-   // if(doEnableExtraLocking)
-   //    elock_fail = extra_lockf(0);
-
     write_lock(&mem->lock);
     while((critical_region_count(current) > 1) && (current->pid > 1) ){ // Wait for now, task is in one or more critical sections, and/or has locks
         nanosleep(&lock_pause, NULL);
@@ -85,9 +81,6 @@ void mem_destroy(struct mem *mem) {
 
     //modify_critical_region_counter(current, -1, __FILE__, __LINE__);
 
-    //if((doEnableExtraLocking) && (!elock_fail))
-    //   extra_unlockf(0);
-
 }
 
 #define PGDIR_TOP(page) ((page) >> 10)

diff --git a/emu/vec.c b/emu/vec.c
@@ -1,6 +1,3 @@
-#ifdef __ARM_NEON__
-#include <arm_neon.h>
-#endif
 #include <math.h>
 #include <string.h>
 
@@ -193,15 +190,8 @@ void vec_imm_shiftrs_d128(NO_CPU, const uint8_t amount, union xmm_reg *dst) {
 }
 
 void vec_add_b128(NO_CPU, union xmm_reg *src, union xmm_reg *dst) {
-#ifdef __ARM_NEON__
-    uint8x16_t neon_dst = vld1q_u8(dst->u8);
-    uint8x16_t neon_src = vld1q_u8(src->u8);
-    uint8x16_t neon_res = vaddq_u8(neon_dst, neon_src);
-    vst1q_u8(dst->u8, neon_res);
-#else
     for (unsigned i = 0; i < array_size(src->u8); i++)
         dst->u8[i] += src->u8[i];
-#endif
 }
 void vec_add_w128(NO_CPU, union xmm_reg *src, union xmm_reg *dst) {
     for (unsigned i = 0; i < array_size(src->u16); i++)
@@ -332,16 +322,9 @@ void vec_xor64(NO_CPU, union mm_reg *src, union mm_reg *dst) {
 }
 
 void vec_min_ub128(NO_CPU, union xmm_reg *src, union xmm_reg *dst) {
-#ifdef __ARM_NEON__
-    uint8x16_t neon_dst = vld1q_u8(dst->u8);
-    uint8x16_t neon_src = vld1q_u8(src->u8);
-    uint8x16_t neon_res = vminq_u8(neon_dst, neon_src);
-    vst1q_u8(dst->u8, neon_res);
-#else
     for (unsigned i = 0; i < array_size(src->u8); i++)
         if (src->u8[i] < dst->u8[i])
             dst->u8[i] = src->u8[i];
-#endif
 }
 void vec_max_ub128(NO_CPU, union xmm_reg *src, union xmm_reg *dst) {
     for (unsigned i = 0; i < array_size(src->u8); i++)
@@ -359,7 +342,7 @@ void vec_maxs_w128(NO_CPU, union xmm_reg *src, union xmm_reg *dst) {
 }
 
 static bool cmpd(double a, double b, int type) {
-    bool res = false;
+    bool res;
     switch (type % 4) {
         case 0: res = a == b; break;
         case 1: res = a < b; break;
@@ -370,7 +353,7 @@ static bool cmpd(double a, double b, int type) {
     return res;
 }
 static bool cmps(float a, float b, int type) {
-    bool res = false;
+    bool res;
     switch (type % 4) {
         case 0: res = a == b; break;
         case 1: res = a < b; break;
@@ -555,9 +538,6 @@ void vec_unpackh_pd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
     dst->f64[0] = dst->f64[1];
     dst->f64[1] = src->f64[1];
 }
-void vec_movlh_ps128(NO_CPU, union xmm_reg *src, union xmm_reg *dst) {
-    dst->qw[1] = src->qw[0];
-}
 
 void vec_packss_w128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
     dst->u32[0] = (satsw(dst->u16[0]) << 0x00) | (satsw(dst->u16[1]) << 0x08) |
@@ -605,15 +585,8 @@ void vec_shuffle_d128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst, uint
 }
 
 void vec_compare_eqb128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
-#ifdef __ARM_NEON__
-    uint8x16_t neon_dst = vld1q_u8(dst->u8);
-    uint8x16_t neon_src = vld1q_u8(src->u8);
-    uint8x16_t neon_res = vceqq_u8(neon_dst, neon_src);
-    vst1q_u8(dst->u8, neon_res);
-#else
     for (unsigned i = 0; i < array_size(src->u8); i++)
         dst->u8[i] = dst->u8[i] == src->u8[i] ? ~0 : 0;
-#endif
 }
 void vec_compare_eqw128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
     for (unsigned i = 0; i < array_size(src->u16); i++)
@@ -624,16 +597,6 @@ void vec_compare_eqd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
         dst->u32[i] = dst->u32[i] == src->u32[i] ? ~0 : 0;
 }
 
-/*
- * Neon algo: (only one part (64bits) is demonstrated, algo works the same for another part)
- * z - is a bit which forms the mask, X - is not interesting bit.
- * neon_src: zXXXXXXXzXXXXXXXzXXXXXXXzXXXXXXXzXXXXXXXzXXXXXXXzXXXXXXXzXXXXXXX...
- * step1:    0000000z0000000z0000000z0000000z0000000z0000000z0000000z0000000z...
- * step2:    00000000000000zz00000000000000zz00000000000000zz00000000000000zz...
- * step3:    0000000000000000000000000000zzzz0000000000000000000000000000zzzz...
- * step4:    00000000000000000000000000000000000000000000000000000000zzzzzzzz...
- * After step4, 8 bits at the end of each 64bit lane are loaded into dst.
- */
 void vec_compares_gtb128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
     for (unsigned i = 0; i < array_size(src->u8); i++)
         dst->u8[i] = (int8_t)dst->u8[i] > (int8_t)src->u8[i] ? ~0 : 0;
@@ -649,19 +612,10 @@ void vec_compares_gtd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
 
 void vec_movmask_b128(NO_CPU, const union xmm_reg *src, uint32_t *dst) {
     *dst = 0;
-#if defined(__ARM_NEON__) && defined(__LITTLE_ENDIAN__)
-    uint8x16_t neon_src = vld1q_u8(src->u8);
-    uint16x8_t step1 = vshrq_n_u8(neon_src, 7);
-    uint32x4_t step2 = vsraq_n_u16(step1, step1, 7);
-    uint64x2_t step3 = vsraq_n_u32(step2, step2, 14);
-    uint16x8_t step4 = vsraq_n_u64(step3, step3, 28);
-    *dst |= (vgetq_lane_u8(step4, 8) << 8) | (vgetq_lane_u8(step4, 0));
-#else
     for (unsigned i = 0; i < array_size(src->u8); i++) {
         if (src->u8[i] & (1 << 7))
             *dst |= 1 << i;
     }
-#endif
 }
 
 void vec_fmovmask_d128(NO_CPU, const union xmm_reg *src, uint32_t *dst) {
@@ -672,12 +626,18 @@ void vec_fmovmask_d128(NO_CPU, const union xmm_reg *src, uint32_t *dst) {
     }
 }
 
-void vec_movl_pd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst) {
-    dst->qw[0] = src->qw[0];
+void vec_movl_p64(NO_CPU, const uint64_t *src, union xmm_reg *dst) {
+    dst->qw[0] = *src;
 }
-void vec_movl_mem_pd128(NO_CPU, const union xmm_reg *src, uint64_t *dst) {
+void vec_movl_pm64(NO_CPU, const union xmm_reg *src, uint64_t *dst) {
     *dst = src->qw[0];
 }
+void vec_movh_p64(NO_CPU, const uint64_t *src, union xmm_reg *dst) {
+    dst->qw[1] = *src;
+}
+void vec_movh_pm64(NO_CPU, const union xmm_reg *src, uint64_t *dst) {
+    *dst = src->qw[1];
+}
 
 void vec_extract_w128(NO_CPU, const union xmm_reg *src, uint32_t *dst, uint8_t index) {
     *dst = src->u16[index % 8];

diff --git a/emu/vec.h b/emu/vec.h
@@ -127,7 +127,7 @@ void vec_unpackh_d128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst);
 void vec_unpackh_dq128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst);
 void vec_unpackh_ps128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst);
 void vec_unpackh_pd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst);
-void vec_movlh_ps128(NO_CPU, union xmm_reg *src, union xmm_reg *dst);
+
 void vec_shuffle_lw128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst, uint8_t encoding);
 void vec_shuffle_hw128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst, uint8_t encoding);
 void vec_shuffle_d128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst, uint8_t encoding);
@@ -141,6 +141,11 @@ void vec_compares_gtd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst);
 
 void vec_movl_pd128(NO_CPU, const union xmm_reg *src, union xmm_reg *dst);
 void vec_movl_mem_pd128(NO_CPU, const union xmm_reg *src, uint64_t *dst);
+void vec_movl_p64(NO_CPU, const uint64_t *src, union xmm_reg *dst);
+void vec_movl_pm64(NO_CPU, const union xmm_reg *src, uint64_t *dst);
+void vec_movh_p64(NO_CPU, const uint64_t *src, union xmm_reg *dst);
+void vec_movh_pm64(NO_CPU, const union xmm_reg *src, uint64_t *dst);
+
 void vec_movmask_b128(NO_CPU, const union xmm_reg *src, uint32_t *dst);
 void vec_fmovmask_d128(NO_CPU, const union xmm_reg *src, uint32_t *dst);
 void vec_extract_w128(NO_CPU, const union xmm_reg *src, uint32_t *dst, uint8_t index);

diff --git a/fs/fake.c b/fs/fake.c
@@ -356,10 +356,10 @@ static void __attribute__((constructor)) init_fake_fdops() {
 
 static int fakefs_mount(struct mount *mount) {
     char db_path[PATH_MAX];
-    strcpy(db_path, mount->source);
+    strncpy(db_path, mount->source, PATH_MAX -1);
     char *basename = strrchr(db_path, '/') + 1;
     assert(strcmp(basename, "data") == 0);
-    strcpy(basename, "meta.db");
+    strncpy(basename, "meta.db", 8);
 
     // do this now so rebuilding can use root_fd
     int err = realfs.mount(mount);

diff --git a/fs/proc/ish.c b/fs/proc/ish.c
@@ -278,18 +278,18 @@ static int proc_ish_show_ips(struct proc_entry *UNUSED(entry), struct proc_data
                 char * int_ip = malloc(100);
                 char * int_dstaddr = malloc(100);
                 if(cursor->ifa_addr->sa_family == AF_INET) {
-                    strcpy(type, "IF_INET");
+                    strncpy(type, "IF_INET", 9);
                 } else {
-                    strcpy(type, "IF_INET6");
+                    strncpy(type, "IF_INET6", 9);
                 }
                 //cursor->ifa_addr->sa_family = AF_INET;
                 get_ip_str(cursor->ifa_addr, int_ip, 100);
                 char * mac = malloc(100);
                 if(cursor->ifa_dstaddr != NULL) {
                     if(cursor->ifa_dstaddr->sa_family == AF_INET) {
-                        strcpy(type, "IF_INET");
+                        strncpy(type, "IF_INET", 9);
                     } else {
-                        strcpy(type, "IF_INET6");
+                        strncpy(type, "IF_INET6", 9);
                         cursor->ifa_dstaddr->sa_family = AF_INET6;
                     }
                     get_ip_str(cursor->ifa_dstaddr, int_dstaddr, 100);
@@ -315,6 +315,7 @@ static int proc_ish_show_ips(struct proc_entry *UNUSED(entry), struct proc_data
                 );
                 free(int_ip);
                 free(int_flags);
+                free(int_dstaddr);
             }
             cursor = cursor->ifa_next;
         }